Enterprise Knowledge Infrastructures, 2nd Ed

Enterprise Knowledge Infrastructures Second Edition Ronald Maier · Thomas H¨adrich · Ren´e Peinl Enterprise Knowledge...

Author: Ronald Maier | Thomas Hädrich and René Peinl

84 downloads 1340 Views 8MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Enterprise Knowledge Infrastructures Second Edition

Ronald Maier · Thomas H¨adrich · Ren´e Peinl

Enterprise Knowledge Infrastructures Second Edition

123

Prof. Dr. Ronald Maier University of Innsbruck Department of Information Systems Universit¨atsstr. 15 6020 Innsbruck Austria [email protected]

Dr. Thomas H¨adrich Open Text Weidestr. 120a 22083 Hamburg Germany [email protected]

Dr. Ren´e Peinl IPI GmbH An der Br¨ucke 4 91586 Lichtenau Germany [email protected]

ISBN 978-3-540-89767-5

e-ISBN 978-3-540-89768-2

DOI 10.1007/978-3-540-89768-2 Library of Congress Control Number: 2009920049 c Springer-Verlag Berlin Heidelberg 2005, 2009 This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the German Copyright Law. The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Cover design: WMX-Design Printed on acid-free paper 9 8 7 6 5 4 3 2 1 springer.com

Preface for the Second Edition

Nearly four years have past by since the first edition of our book was published and we are very happy that the book has found its market and even more about the feedback from those readers who took some time to share their opinions about the first edition with us. As time has gone by, a lot has happened both in the development and adoption of technologies powering enterprise knowledge infrastructures and in the professional and personal lives of the authors. After our joint time at Martin-Luther-University of Halle-Wittenberg, life paths of the authors split again, so that our team turned into a virtual one, geographically spread from Hamburg via Hof to Innsbruck. Co-authoring this edition provided us with numerous opportunities to reinforce our excitement about how convenient it is to collaborate across locations. We have gained plenty of new experiences in research as well as in practice and several of them have been included in this new edition. One surprising experience is that despite all changes in information technologies, like the advent of Web 2.0 and the corresponding acceleration of browser-based applications as well as the explosion of user-generated contents, the core proposition of the book still remains an attractive and challenging one. The need for integration of existing systems instead of introduction of new isolated ones is unchanged. It typically takes many years until new technologies are adopted, particularly in traditional businesses and organizations. There are many reasons for this phenomenon which has been profoundly described in numerous technologyadoption models in the literature. The majority of enterprises does not qualify as IT innovators and thus refrains from investing into what could be a hype without prove that there are measurable benefits to be gained. Integration efforts are especially challenging endeavors because they are often spanning organizational units and need sponsors and champions that think in a more holistic way instead of single system classes. Not only technical boundaries have to be bridged, but also organizational boundaries with people fearful about loosing power over their system if it is integrated into the overall infrastructure. So, enterprise knowledge infrastructures can be seen as complementing organizational efforts to bridge departmental boundaries with process-orientation, in this case knowledge processes, not business processes as with ERP solutions. The second edition reflects this understanding of enterprise knowledge infrastructures by rearranging knowledge services into the 5-I model of knowledge maturing. This new substructure of the book is intended to allow for better teaching and learning. It takes on a more dynamic, process-oriented perspective on knowledge and knowledge services. The authors hope that the major additions to knowledge services make the vision of the book more comprehensible. The edition has been profoundly extended and completely revised throughout all chapters. Product examples and overviews have been updated to the latest market data. Updated definitions and conceptual foundations ease understanding of the subject matter. Case examples have been added for each main chapter and, last but not least, the general didactic approach of the book has been improved. Such an effort in addition to three full-time jobs is only possible with numerous helping hands for proof-reading, suggestions for further improvements and taking over some other

VI

Preface for the Second Edition

tasks unrelated with the book in order to free time which we could invest in thinking, discussing and writing. The latter is especially true for our partners who have been most understanding about us indulging into writing a book which cannot be taken for granted at all. We also like to thank our students and colleagues, the team at the Dept. of Information Systems, School of Business at the University of Innsbruck, and especially Nadine Amende, Dominik Grimm, Andreas Kaschig, Tanja Kohn, Alexander Sandow and Stefan Thalmann. Some of them are former students who have decided to take on an academic career and have expanded our social network to new universities and renewed the links to colleagues with established relationships. We find it essential to keep exchanging opinions, especially with those having different professional backgrounds as they bring in new ideas and different views. We hope that you, the readers, find the book inspiring for further exploring ways how IT can increase productivity of knowledge work in the 21st century. Finally, we would like to invite you to share your opinion about our book and the ideas behind most easily by email ([email protected], [email protected], [email protected]).

Ronald Maier Thomas Hädrich René Peinl Innsbruck, Hamburg, Hof and many places with Internet connections, November 2008

Preface for the First Edition

Both, academics and practitioners alike have spent considerable efforts during the last years to establish ICT support for the handling of knowledge, an idea that is almost as old as the field of computer science. Not surprisingly, the solution is still not there and many businesses trying to implement these technologies have been frustrated by the fact that the technologies certainly could not live up to the overly high expectations. However, there are still numerous projects in organizations that try to tackle the fundamental challenge of how to increase productivity of knowledge work. People do not believe in quick solutions to this problem any more - and they are right. Knowledge management is dead. Long live knowledge management! Central hypothesis of this book is that the implementation of KM technology in organizations has entered a new stage. In the last years, many vendors jumped on the bandwagon and insisted that their products had “knowledge management technology inside”. More recently, however, it seems that many technologies provided by avantgarde systems to support handling of (documented) knowledge, finding of, collaboration between and learning by people doing knowledge work, were weaved into the enterprise infrastructure implemented in many organizations. It is not anymore the quest for the best individual tool targeting a specific KM problem that organizations should engage in. Organizations should strive for improving their information and communication infrastructures so that they are able to handle semantic descriptions of integrated, semi-structured data and offer advanced knowledge services on top of them. Within this field, the book combines a thorough treatment of the vision of an ideal enterprise knowledge infrastructure on the one hand with a comprehensive description of concepts, standards, tools and systems that are already available and can help to implement this vision on the other hand. We hope that the book will help you to understand the complex matter, that you will enjoy the ideas presented here, discuss them in teams and communities, gain new insights by answering the questions and exercises and that you will be motivated to develop them further. Additional support and contents can be found at the supporting Web site. Please visit: URL: http://www.wiwi.uni-halle.de/maier/EKI/. You can also contact us by email. Any comments are most welcome at: [maier, haedrich, peinl]@wiwi.uni-halle.de! The book presents the results of the development of courses and programs in knowledge management (systems) for the University of Regensburg, Danube-University of Krems, Austria, and Martin-Luther-University of Halle-Wittenberg. In the last two years, the authors have jointly developed five courses that together present and train to use the concepts in this book at the Department of Management Information Systems, Information Systems Leadership of the Martin-Luther-University of Halle-Wittenberg. During this period we also established a basic knowledge infrastructure at our Department that helped us to exchange ideas and step-by-step develop the concepts that are now part of this book. Many people have contributed to our thoughts on enterprise knowledge infrastructures. We would like to thank our students for sharing their experiences gained in many organizations implementing KM technologies who inspired us to come up with a book that consis-

VIII

Preface for the First Edition

tently presents the material scattered across a large number of sources, our teaching assistants for drawing some of the figures, for their support in implementing some of the tools and for numerous remarks on the material and last but not least all colleagues at our Department as well as our friends working in other Departments spread all over the world for many fruitful discussions and proofreading of the manuscript. Ronald Maier Thomas Hädrich René Peinl Halle / Saale, February 2005

Contents

Preface for the Second Edition Preface for the First Edition 1 Foundation

.....................V . . . . . . . . . . . . . . . . . . . . . VII

..................................... 1

1.1 Knowledge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.1.1 Knowledge in Organizational Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.1.2 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 1.2 Knowledge Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 1.2.1 Definition and Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 1.2.2 Traditional Work versus Knowledge Work . . . . . . . . . . . . . . . . . . . . . . 24 1.3 Knowledge Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 1.3.1 Roots of Knowledge Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 1.3.2 From Data to Knowledge Management . . . . . . . . . . . . . . . . . . . . . . . . . 29 1.3.3 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 1.4 Knowledge Management Instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 1.4.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 1.4.2 Product-oriented Instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 1.4.3 Process-oriented Instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 1.4.4 Knowledge Maturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 1.5 Information System Architectures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 1.5.1 System Architectures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 1.5.2 Enterprise Architectures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 1.6 Knowledge Infrastructures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 1.6.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 1.6.2 Knowledge Management Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 1.6.3 Architecture of Knowledge Infrastructures . . . . . . . . . . . . . . . . . . . . . . . 75

2 Infrastructure

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

2.1 Network Infrastructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1 Topologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.2 Geographical Expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.3 Layered Network Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

84 85 88 89

X

Contents

2.2 Network Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 2.2.1 Physical and Data Link Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 2.2.2 Network and Transport Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 2.2.3 Application Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 2.2.4 Network Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 2.2.5 Network Hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 2.3 Infrastructure Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 2.3.1 Storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 2.3.2 Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126 2.3.3 Application Infrastructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

3 Integration Services

. . . . . . . . . . . . . . . . . . . . . . . . . . . 151

3.1 Data Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.1 Semi-structured Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.2 Uniform Resource Identifier and Unicode . . . . . . . . . . . . . . . . . . . . . . 3.1.3 Extensible Markup Language (XML) . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.4 XML Schema . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.5 XSL Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Meta-data-based Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1 Meta-data Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2 Resource Description Framework (RDF) . . . . . . . . . . . . . . . . . . . . . . . 3.2.3 Storing and Querying Meta-data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.4 Meta-data Dimensions ..................................... 3.3 Ontology-based Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1 Ontology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2 RDF Schema . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.3 Web Ontology Language (OWL) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.4 Higher Levels of the Semantic Web Stack . . . . . . . . . . . . . . . . . . . . . . 3.4 Function and Process Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.1 Function Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.2 Process Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.3 Web Service Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.4 Semantic Web Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4 Knowledge Services

153 154 156 157 162 165 172 175 178 183 191 205 206 210 213 218 219 219 225 231 234

. . . . . . . . . . . . . . . . . . . . . . . . . . . . 241

4.1 Investigation Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.1 Foundation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.2 Exploration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.3 Search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.4 Presentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.5 Monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

242 244 246 257 264 267

Contents

4.2 Individuation Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1 Foundation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2 Experience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.3 Expression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.4 Competence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.5 Expertise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Interaction Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.1 Foundation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.2 Initiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.3 Networking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.4 Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.5 Community . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 In-form-ation Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.1 Foundation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.2 Capturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.3 Manipulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.4 Translation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.5 Archiving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5 Instruction Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5.1 Foundation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5.2 Enriching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5.3 Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5.4 Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5.5 Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5 Access Services

XI

271 272 273 276 278 281 285 286 290 292 294 301 305 306 312 318 321 326 331 333 341 346 353 357

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365

5.1 Server-side Access Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.1 Portals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.2 Personalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Client-side Access Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.1 Thin Clients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.2 Rich Thin Clients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.3 Rich Clients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.4 Desktop Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Mobile Access . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.1 Mobile Computing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.2 Mobile Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.3 Mobile Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

366 366 370 374 374 378 380 382 390 390 394 399

XII

Contents

6 Conclusion 6.1 6.2 6.3 6.4 6.5 6.6

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407

Situation-oriented Support of Knowledge Work . . . . . . . . . . . . . . . . . . . . . Context Factors for Knowledge Sharing . . . . . . . . . . . . . . . . . . . . . . . . . . . State-of-Practice of Knowledge Management . . . . . . . . . . . . . . . . . . . . . . . Centralized versus Distributed Architectures . . . . . . . . . . . . . . . . . . . . . . . Chances and Risks of Web 2.0 from a Knowledge Perspective . . . . . . . . . The Future of Knowledge Infrastructures . . . . . . . . . . . . . . . . . . . . . . . . . .

Bibliography Index

408 412 416 420 424 429

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 439

1 Foundation

Information and communication technologies (ICTs) to support handling of knowledge in organizations have been discussed for quite a long time. In the beginning of applying ICTs, it was routine work that was first prone to automation. ICTs gradually broadened their scope and have been extended to embrace highly valued work that years before hardly anybody would have believed that it could be supported by ICTs. Back in the 50s to 80s of the last century, various waves of systems applying artificial intelligence (AI) technologies had a powerful impact on concepts of knowledge, not only in the discipline computer science, but also in fields, such as cognitive psychology. However, many business organizations trying to implement these technologies, first advertised as “general problem solver”, were frustrated by the fact that the technologies could not live up to the overly high expectations. Instead, they showed comparably high complexity and difficulties in applying them to business challenges. Thus, AI technologies survived only in special and narrow application fields. In the 90s, after a period of high attention to increasing efficiency, organizations were faced with the transformation of society into a knowledge society, of economy into a knowledge economy and its challenges to significantly increase the speed of innovation and improve the way organizations handle distributed and fragmented knowledge. For those countries that have not (any more) the possibility to exploit some form of natural resources, it is primarily or even only knowledge that creates wealth. Knowledge work requires a high level of skills and expertise from employees and an organizational design that creates an environment conducive for this type of work. Increasing productivity of knowledge work is topical in a time of increasing knowledge intensity of processes in businesses and organizations. It requires continuous and consequent commitment from all organizational levels. Concepts of knowledge management (KM) have been suggested to meet these challenges, starting with highly innovative work by authors such as Davenport, Nonaka, Sveiby or Wiig, just to name a few. Many authors from a variety of disciplines have created, applied and reflected a number of approaches, concepts, methods, tools and strategies for knowledge management. In its short history, knowledge management has absorbed a wide array of research questions which has made it interesting and attractive for a large community as diverse as its authors with backgrounds in management science, organization science, psychology or computer science. At the same time, however, the field of knowledge management has struggled with numerous terms used differently, incommensurable approaches and its lack of clear profit impact in a business context.

Knowledge

Knowledge work

Knowledge management

2

1. Foundation

Knowledge management phases

During the last twenty years, businesses have faced four distinctive phases of KM. The first phase could be termed human-oriented KM. Organizations realized the value of their “human capital” and bundled a number of instruments aiming at the individual knowledge worker and her productivity. The next phase was backed by tremendously increased opportunities offered by ICTs and could be called technology-oriented KM. Organizations were eagerly experimenting with new ICTs in attempts to benefit from the promised changes that would come about by implementing KM tools and systems. In a third phase which primarily was fueled by the emphasis on business processes typical for German-speaking countries, KM methods, tools and instruments were repositioned as knowledge processes and linked to knowledge-intensive business processes. Thus, KM initiatives could be designed with the same language as used in organizational design and IT support of business activities in general, the language of business processes. After human-oriented, technology-oriented and process-oriented KM, recently a fourth KM phase has reached businesses backed by the hype keywords Web 2.0 and social software: collaborative KM. While in many organizations knowledge workers are busy trying out new alternatives for production of contents, for networking and for selfdirected learning, questions arise how these activities can be coordinated or guided so that they are in line with organizational goals. Backed by tremendous interest in KM in academia and business practice, vendors of ICT systems as well as researchers showed prototypes, tools and systems to support KM called knowledge management systems (KMSs). This term, however, is a misnomer. On the one hand, knowledge in many definitions (section 1.1.2, 15ff) is either bound to people or extracted from an expert and made available in specially designed systems, so-called knowledge-based systems. On the other hand, management is a term that denotes the software-supported handling, e.g., storing, administering, updating and retrieving of (business) objects when used in connection with ICTs. Examples are database management systems or document management systems. However, strictly speaking, KMSs neither contain knowledge nor do they manage it. The term KMS has been a strong metaphor for developing a new breed of ICT systems, though. In this view, KMS combine, integrate and extend a number of heterogeneous ICTs, e.g., AI, communication, coordination and collaboration systems, content, document and learning (content) management systems, search and visualization systems. Given the complexity of these technologies, it seems obvious that the development of KMSs is a complex undertaking. In the last years, many vendors have insisted that their products have “knowledge management technology inside”. More recently, however, it seems that many technologies that have been used by employees to support knowledge work have been woven into the ICT infrastructure implemented in many organizations. Whereas enterprise resource planning sys-

Knowledge management systems

Knowledge infrastructure

1.1. Knowledge

tems target the informational representation of business transactions, enterprise knowledge infrastructures (EKIs) create an ICT environment for knowledge work throughout the organization. Chapter 1 provides the foundation for the numerous technologies that can be used to build EKIs. Section 1.1 discusses the term knowledge and distills specifics that are required to understand its use in connection with the terms management, work and infrastructure. Section 1.2 reflects on the underlying characteristics of the type of work that has to be supported by EKIs called knowledge work. Section 1.3 discusses the most important approaches and concepts of knowledge management. Section 1.4 specifically targets instruments that have emerged from 20 years of research on knowledge management and can more or less readily be applied in organizations. Finally, section 1.6 introduces the key term in this book, EKI. On completion of this chapter, you should be able to define the most important KM concepts and approaches, identify the many facets that the term knowledge has in different perspectives and analyze the challenges for its systematic management, employ the presented framework to classify knowledge along a number of important dimensions, analyze the potentials of KM in organizations, appreciate the need for a systematic handling of knowledge to improve productivity of knowledge work, identify the changed requirements for the design of ICT posed by knowledge work, describe state-of-the-art KM instruments applicable in organizations, define the concept of architecture and discuss its benefits, distinguish types and alternatives of architectures in organizations, define the concept of (Web) service that is at the core of a service-oriented architecture, explain the EKI concept and relate it to the broader concept of an organization’s information and communication landscape, identify EKI layers and services.

3

Overview

Learning objectives

1.1 Knowledge The importance of knowledge for societies and particularly organizations is rarely questioned and has been studied for a long time. The foundation for Western thinking about knowledge can be traced back to Greek philosophy. However, this book cannot give a comprehensive overview of defi-

Roots in philosophy

4

Organizational perspective

1. Foundation

nitions of knowledge because even a limited review of work done in philosophy would fill books, nor can it give an all-encompassing definition. Instead, some important conceptualizations of knowledge which have made their way into various classes of KM approaches will be reviewed from an organizational perspective (section 1.1.1). Due to the major role that organizational knowledge plays, there are a number of related terms that have to be clarified, such as capability, competence, expertise or intellectual capital. Some facets of the term knowledge will be selected to discuss the implications on the definition, the design and the implementation of EKI. Then, the term is defined in the context of EKIs (section 1.1.2). Also, important dimensions will be distinguished that help to classify knowledge used in organizations.

1.1.1 Knowledge in Organizational Settings Knowledge transforms organizations

Use of the term knowledge

Data

Information

The transformation of organizations into knowledge-intensive and knowledge-aware organizations takes place at an ever-increasing pace. Knowledge as the key resource, not labor, raw material or capital, changes production functions in organizations significantly. Knowledge represents the key concept to explain the increasing velocity of the transformation of social life in general and the way businesses and social institutions work (Drucker 1994). The term knowledge is used widely, but often quite vaguely, even within the KM field. There are many definitions which differ not only between scientific disciplines contributing to KM, but also within these disciplines and thus also within the KM field. Moreover, the definitions of knowledge lead to different perspectives on organizational knowledge and, thus, to different concepts of interventions into an organization’s way of handling knowledge. Relation to other concepts. Knowledge is related to many other concepts. The most often cited relationships are those to data and information. Figure 1-1 shows a common depiction of the relationships between data, information and knowledge. Data refers to symbols, e.g., characters or numbers, ordered to an elementary description of a person, thing, event, activity, transaction or state in the perceived reality or imagination of persons. Data can be recorded, classified and stored, but are not organized to convey any specific meaning. Data items can be numeric, alphanumeric, figures, sounds or images. With respect to ICT, data items are stored in organized databases. Information is seen in a multitude of ways, but most definitions draw the line between data and information with respect to meaning, the semantics that are commonly assigned to interpreted data, also called information, but not to (raw) data. There are basically two main perspectives:

1.1. Knowledge

5

Information is data that have been organized so that they have meaning and value to the recipient. The recipient interprets the meaning and draws conclusions and implications. Information is the result of a person’s interpretation of signals from the environment, whereby the result depends on the person’s knowledge and the context of the interpretation. pragmatics

knowledge

relation to user/ application

link, network with context / experiences semantics

information

relation to meaning

interpret syntactics

sigmatics

Figure 1-1.

data

relation to other symbols

structure, order, format relation to denominated symbol object / activity

Data, information and knowledge as hierarchy of terms

Many classifications of the term knowledge use a dichotomy to describe one type of knowledge and its opposite. Table 1-1 presents some examples for important classes of knowledge that are organized with respect to person, organization, context and ICT (Maier 2007). The variety of definitions of the term knowledge is due to the variety of research subjects which require more or less focus on knowledge. At least to some extent, there is agreement among KM researchers about the most important dichotomies and characteristics of knowledge, such as individual versus organizational, implicit versus explicit, organization-internal versus organization-external knowledge (section 1.1.2). Consequences for KM. In the following, some important characteristics of knowledge are summarized which have consequences or provide challenges for EKI design: Unlike information, knowledge is not easily transferred. The costs for the “distribution” of knowledge can be very high. Unlike information transfer, it takes time for individuals to reconstruct knowledge because this process not only requires interpretation as in the case of information, but also requires learning. Thus, it takes substantially more effort to implement a systematic management of knowledge transfer as compared to the transfer of information. There are a number of institutions that provide an

Classes of knowledge

Agreement about important dimensions

Economic differences to information

6

1. Foundation

environment conducive to knowledge transfer and thus to learning. This environment can be viewed as an activity system in which “knowledge seekers”, “students” or “apprentices” not only directly learn from “knowledge providers”, “teachers” or “masters”, but also from participating in a community of practice of all the knowledge seekers and knowledge providers in a joint setting, e.g., schools, universities, management centers, corporate universities, industry organizations offering apprenticeships. Table 1-1. area

dimension

values

context

abstraction

narrative/concrete - scientific/abstract

generalization

particular - universal

representation

declarative - procedural

access

accessible - inaccessible

medium

electronic - non-electronic

codability

codable - non-codable

relevance

relevant - irrelevant

authorization

informal - formal

security

private - public

ownership

internal - external

value

valuable - not valuable

awareness

implicit/tacit - articulated/explicit

support

supported/dominant - unsupported/minority

existence

knowledge - not knowledge

ICT

organization

person

Figure 1-2. “Transfer” of knowledge

Transfer of data vs. transfer of knowledge

Classification of knowledge

Classification of knowledge

Several authors dealing with ICT support for KM have written about systems supporting transfer or distribution of knowledge. In this area, not only explicit knowledge is considered which can be transferred with the help of, e.g., documents, but also the tacit side of knowledge. The latter can only be handed on directly from teacher to learner (socialization). According to most definitions of data, information and knowledge only data can be transported or communicated which in turn is interpreted by individuals or social systems. Therefore, even knowledge infrastructures essentially contain and support communication of data, not knowledge. However, the “transfer” or “sharing” of knowledge denotes the simplified and shortened process including interpretation of the message (information) and actualization or extension of knowledge by the receiving system.

1.1. Knowledge

7

Figure 1-3 shows the complete communication process of data, information and knowledge. Transfer of knowledge implies that the sender is quite certain that the receiver will be capable of interpreting the data accordingly, (re-) construct the knowledge and use it to actualize the receiver’s knowledge in a way that the sender intends. system A

system B

knowledge

interpretation

information directs attention

directs attention

actualizes, extends

information directs attention

get signals

activity system

sensors

data

Figure 1-3.

guides & limits

interpretation

get signals

sensors

knowledge purpose for action

guides & limits

purpose for action

directs attention

actualizes, extends

activity system

channel

The transfer of information and knowledge

Note that the sender cannot be sure that the receiver will interpret the data as intended by the sender. Additionally, according to modern theories in the cognitive sciences with each transfer of knowledge, the knowledge itself is changed not only at the receiving end, but also at the sending end of the communication as it is not just “retrieved” in memory, but reconstructed and the knowledge’s context is thus changed with each transfer. Knowledge is developed in a cultural context with social, political, economic and ideological dimensions that exert continual forces on both the substance and the process of scientific knowledge creation. What has been said about scientific knowledge creation is all the more true in organizational settings. Organizations are not regularly striving for absolute truth, but for a socially constructed reality that allows for successful organizational actions. Knowledge cannot be separated easily from the context of its creation, reception and application. Another important challenge in organizations is protecting valuable knowledge, e.g., against industrial espionage. Examples for measures that prevent the unwanted use of organizational knowledge are classification or property laws and organizational instruments such as incentives, conduct rules or postponing of rewards because a great deal of knowledge valuable to an organization resides with single employees. In some cases, it is opportune for organizations to share knowledge with competition (coopetition) and thus systematically manage diffusion of otherwise restricted, i.e. patented, classified or confidential knowledge,

Reconstructing knowledge

Relation to context

Protection of knowledge

8

Knowledge as product vs. process

“Right” quantity of knowledge

Multi-faceted knowledge

Role of knowledge in different types of organizations

1. Foundation

e.g., through mechanisms such as visiting each other’s production facilities, consortia or benchmarking. One implication on the design of EKI is that valuable knowledge must be protected from leaving the organization unintentionally, so it might not be appropriate to make it completely transparent, e.g., to publish it on the organization’s Intranet, but to disaggregate the knowledge so that it cannot be taken easily to a competitor. Knowledge can be conceptualized as a product or as a process. Both concepts are important, though they have differing implications on the design of EKI. Basically, explicit knowledge can be documented and stored in knowledge repositories whereas (more) implicit knowledge has to be supported indirectly through ICTs used to broker and handle communications. Many KM approaches implicitly hold the presupposition that the more knowledge an organization holds, the better for the organization. Applying this simple equation can be dangerous because it does not consider, e.g., that knowledge created in an organization might not be useful, that communicating knowledge expects quite a lot from the receiving system (individual or social), namely that the system rebuilds its knowledge structures, that knowledge is in a sense provisional and is held until better knowledge is created, that more measurable knowledge in terms of, e.g., publications or documents not necessarily means that the organization can act or interpret more intelligently, that knowledge increases “not knowledge” which causes the paradox that the more an organization knows, the more knowledge it demands which in turn leads to less efficient daily operations. As a consequence, EKI have to consider this danger of information overload and inefficient “oversupply” of knowledge. Attention has to be paid to, e.g., contextualization, filtering, profiling and to determining the optimal portion, level and granularity of knowledge that is presented. EKIs differ in design and implementation from more traditional application systems. The term knowledge as used here comprises among others valuations, opinions or forecasts, whereas more traditional application systems focus more or less exclusively on hard data. Also, the design of EKI has to consider the multiple electronically available sources of data such as documents, files, messages, contributions in newsgroups, multimedia elements or links to these sources which all might contain useful knowledge once structured, linked and contextualized. Thus, EKI can be combined with an organization’s already existing information systems (IS). Classifications of knowledge can be used to postulate different requirements or perspectives for KM initiatives and supporting ICTs. Table 1-2 shows four types of organizations that differ with respect to the focussed type of knowledge and thus require support by different ICT.

1.1. Knowledge

9

The distinction uses the organizational level from which the primary contributions to the fulfilment of organizational goals is expected, i.e. individual versus collective, and whether the focus is on familiar or on novel problems. Empirical analysis suggests trends that organizations are transformed from type I, II and III into type IV organizations. Table 1-2.

Organizations according to types of knowledge (Blackler 1995, 1030) Type I: expertdependent

Type II: knowledge-routinized

Type III: symbolic- Type IV: commuanalyst-dependent nication-intensive

level

focus on individual

focus on collective

focus on individual

focus on collective

type of problems

familiar problems

familiar problems

novel problems

novel problems

type of embodied compeknowledge tencies of key members

embrained skills of knowledge embedded in technolo- key members gies, rules and procedures

characterization

performance of specialist experts is crucial; status and power from professional reputation

capital, technology or labor-intensive; hierarchical division of labor and control

entrepreneurial problem solving; status and power from creative achievements

key processes: communication, collaboration, empowerment through integration

example

professional bureaucracy, e.g., hospital

machine bureaucracy, e.g., traditional factory

knowledge-intensive firm, e.g., software house

adhocracy, innovation-mediated production

computer integrated work systems

information support and XPS design

development of cooperation systems (CSCW)

role of ICT computer displacement of action skills

encultured knowledge and collective understanding

In the context of management science, concepts can be distinguished that stress the importance of knowledge as a strategic organizational resource. It is well worth to briefly review these concepts and their theoretical basis because the distinctive definitions of knowledge and related concepts help to understand the different perspectives taken in the literature and also allow for a characterization of KM approaches. This perspective is called the resource-based view and builds on ideas presented in the theory of the growth of the firm (Penrose 1959, Wernerfelt 1984). Central idea of the resource-based view is that an organization’s success is determined by the existence of organization-specific unique resources. As opposed to the market-based view (Learned et al. 1965, Porter 1980), competitive advantages are not due to superior positioning of an organization in an industry, but due to superior quality of resources or a superior use of the organizational resources. Heterogeneity of resources between organizations enables sustained competitive advan-

Strategic aspects of knowledge

Resourcebased view

10

Enduring resources

Knowledgebased view Terms stressing strategic relevance

Classification of resources

Classes overlap

1. Foundation

tages and is determined by individual historic developments of organizations, by developing specific material and immaterial resources and by creating complex organizational routines which cause specific historical trajectories and lead to unique idiosyncratic combinations of resources. Another central assumption of the resource-based view is that in uncertain and dynamic competitive environments, products and services demanded in the market change quickly, whereas resources and capabilities are more enduring. As a consequence, proponents of the resourcebased view suggest to base a strategy on resources rather than on productmarket combinations as suggested in the market-based view. Resources are seen as platforms for developing varying products and services. A more focused version of the resource-based view is called the knowledge-based view, stresses the importance of knowledge as an organizational resource and plays a role in embedding KM into corporate strategy. In order to avoid confusion with the traditional view on the term resource and stress the strategic relevance of organization-internal assets, several terms have been proposed which are discussed in the following. Organizational resource. A firm’s resources at a given time could be defined as those (tangible and intangible) assets which are tied semi-permanently to the firm. This organization-specific element is what distinguishes resources in the resource-based view from the traditional viewpoint in economics or business administration with its primary production factors land, labor and capital. Resources in the resource-based view typically have to be built and cannot be bought. Organization-specific resources can be classified in a multitude of ways similar to knowledge (see Figure 1-2 on page 6). Figure 1-4 presents a typical classification of resources and gives some examples. Tangible resources are detailed in financial and physical resources. Intangible resources are classified into person-dependent and person-independent ones. Person-independent resources are further divided into intangible and organizational assets. Intangible assets have a relationship to the organization’s environment because they are either legally secured, e.g., patents, intellectual property, or refer to business partners, e.g., networks, customer relationships, reputation. Organizational assets refer to the organization’s culture, e.g., willingness to share knowledge, perception of service and quality, and routines, e.g., learning cycles, managerial systems, and do not have a direct relationship to the organization’s environment. The detailed classes overlap to some extent, especially with respect to the dimension person-dependency as, e.g., smooth functioning of networks, classified here as person-independent, certainly depends on contacts of single employees.

- contextual ICT (e.g. KMS) - specific production facilities

physical resources

contracts patents licences intellectual property trade secrets brands

-

- reputation - networks - customer and supplier relationships

positional

within a legal context

differential

without a legal context

regulatory

- borrowing capacity - internal funds generation

differential

financial resources

intangible assets

organizational routines - production processes - flexible workflows - continuous process improvement - learning cycles - managerial systems

- willingness to share knowledge - perception of quality - ability to manage change - perception of service

organizational assets

intangible resources

organizational culture

person-independent resources

competitive advantages

cultural

tangible resources

differential

Classification of resources (Maier 2007, 99) quality differential

Figure 1-4. organization/managerial

organization-specific resources

- expert knowledge - creativity - non-explicable know-how

tacit knowledge

- explicable individual knowledge - skills

explicit, personal knowledge

person-dependent resources

1.1. Knowledge 11

12

Capability differentials

Depreciation of knowledge

Characteristics of strategic resources

1. Foundation

Organizational capabilities. Resources are combined, consolidated or applied in an organizational context to form capabilities which are “teams” of resources working together or an interconnected set of knowledge collections in the sense of a tightly coupled system. In situations of quickly changing complex environments, dynamic capabilities are crucial which are defined as a firm’s ability to integrate, build, and reconfigure internal and external competencies to address rapidly changing environments. Figure 1-4 also shows that the value of sets of interconnected organizational resources has to be determined in relation to the competition. A comparison reveals so-called capability differentials. Five types of capability differentials can be distinguished: functional/business system differentials: result from knowledge, skills and experience of employees and others in the value chain, e.g., suppliers, distributors, lawyers, agents working for the organization, cultural differentials: apply to the organizational culture as a whole; however, organizational routines are considered as functional differentials because they are transparent and subject to systematic and intended change as opposed to the organizational culture. Cultural differentials are also closely related to organization differentials, organization or managerial quality differentials: result from an organization’s ability to consistently innovate and adapt more quickly and effectively than its competitors. As it is probably easier to systematically influence the quality of managerial systems than it is to influence the organizational culture, managerial systems might constitute a factor that can be distinguished from cultural differentials, positional differentials: are a consequence of past actions which build reputation with business partners, especially customers, regulatory/legal differentials: result from governments limiting competitors to perform certain activities. Regulatory differentials thus are based on those resources that are legally secured, such as patents, contracts, licences or trade secrets. To sum up, resources are the basis for capability differentials. Capability differentials provide competitive advantages which can be leveraged in order to produce superior products and services. As organizational capabilities are determined with respect to competitors, it is only a certain time frame during which they provide competitive advantages. The speed with which innovations penetrate entire industries creates the need to reduce the cycle-time with which new processes and technologies are applied and turned into new products or services. Strategic capabilities or (core) competencies. Capabilities and competencies are often used synonymously. Organizational competencies, however, are focussed on knowledge as resource, therefore on the functional/ business system differential. They are directly related to an organization’s

1.1. Knowledge

strategic choices and are based on a combination or integration of the (individual and common or organizational) knowledge. In order to be strategically relevant and capable of providing sustained competitive advantages, resources must have the following characteristics: (1) Resources must be rare, otherwise competitors can access them easily and no competitive advantage can be gained from their use. (2) Resources must either enable organizations to create value for their customers, thus contributing significantly to the perceived customer benefits or to substantially improve effectiveness and efficiency of the organization’s processes. Additionally, the value of a resource depends on the relative advantage it bears when compared to the competition. (3) Resources must provide potential access to a variety of markets. In other words, resources must be applicable in a multitude of products and services and a multitude of markets in order to be of strategic relevance. (4) Resources must not be easily replicated in rival organizations. Replication is difficult, e.g., due to unique historical conditions in creating the resources, causal ambiguity, i.e. imperfect information or lack of transparency, social complexity, i.e. several individuals jointly provide competitive advantages, or embedding in organizations, i.e. several resources can be complexly interrelated and integrated within an organization’s routines or culture. Thus, there exist so-called barriers to imitation in analogy to the entry or mobility barriers in the market-based view. (5) Resources must not be easily substituted by other resources in order to generate sustained competitive advantages. Thus, a variety of other resources can threaten strategic relevance of a resource. (6) A competitive advantage will be the more sustained, the more difficult it is to acquire the resource, e.g., in cooperation with other organizations. The reasons for a lack of transferability are partly the same as the ones presented for lack of imitability, e.g., geographical immobility, imperfect information or the fact that resources are firm-specific. (7) Longevity of competitive advantages depends upon the rate at which the underlying resources depreciate or become obsolete. Durability varies considerably, e.g., technological resources depreciate quickly due to the increasing pace of technological change whereas reputation and brands are more durable. (8) Resources must be legally undisputed. Profits from a resource can be subject to bargaining, e.g., with business partners, such as customers, suppliers or distributors, and employees. The more knowledge work is on the rise, the more employees know of their capabilities and negotiate with their employers about the value of their contributions. The more an employee’s contribution is clearly identifiable, the more mobile this employee is. The easier an employee’s contributions to capabilities can be transferred to other organizations, the stronger is the employee’s position in negotiations with the organization.

13

Scarce Competitively superior

Multi-purposeful Non- or imperfectly imitable

Non-substitutable Non-transferable

Durable

Appropriable

14

1. Foundation

From resources to competitive advantage

Figure 1-5 depicts a framework showing the chain of arguments used in the resource-based view for conceptualizing the relationship between resources and the more recent concepts of organizational capabilities or competencies and in turn their relationship with competitive advantages. Consequent management of organizational resources thus has to identify, select, develop, synergistically connect, transform and retain organizational resources and integrate them into strategically relevant capabilities.

4. Select a strategy which best exploits the firm’s resources and capabilities relative to external opportunities.

3. Appraise the rent-generating potential of resources and capabilities. They need to be:

competitive advantage

industry factors

5. Identify resource gaps. strategy

(a) scarce, superior, durable, multipurposeful

Invest in replenishing, augmenting and upgrading the firm’s resource base.

(b) non-imitable, non-substitutable, non-transferable, appropriable 2. Identify the firm’s capabilities: What can the firm do more effictively than its rivals? Identify the resource inputs to each capability, and the complexity of each capability.

1. Identify and classify the firm’s resources. Appraise strenghts and weaknesses relative to competitors. Identify opportunities for better utilization of resources.

Figure 1-5.

organizational capabilities

resources

Relationship between resources, capabilities, competitive advantages and strategy (Grant 1991, 115, 1998, 113)

According to the knowledge-based view, competitive advantage of an organization depends on how successful it is in exploiting, applying and integrating its existing capabilities and in exploring and building new capabilities that can be applied to the market. Expertise. Experiences or expertise are concepts that refer to an individual’s knowledge base rather than the organizational connotation of terms such as capability or competence. Research on expertise has a long tradition in psychology and pedagogy. Results show to what extent problem solving or information processing by experts differs from novices or intermediates. Expertise relates to a domain in which the expert intuitively applies highly inter-connected, or encapsulated knowledge in order to interpret situations and propose solutions. Expertise is grounded on longstanding experience in the domain, many sources speak of ten years of continuing experiences needed although this view has been challenged

1.1. Knowledge

15

with respect to the fast-moving domain of IT. Experts are often widely recognized as a reliable source of knowledge. This can even have legal implications in that employees can officially back their actions by an expert’s opinion or expertise, used here as a homonym in the sense of a written account about a certain topic, without having to fear legal prosecution. Experts are critical resources in organizations so that their contributions need to be carefully handled. Also, experts supposedly require different support by EKIs than novices. The challenges in designing EKIs that consider experts in organizations will be discussed in section 4.2.5, 281ff. Intellectual Capital. One of the most prevalent questions in the KM area widely discussed in literature and practice is how to determine the value created and the benefits gained by such initiatives. Apart from traditional measures for firm performance such as ROI, ROA, ROE or EVA, several approaches for measuring performance in KM can be distinguished, e.g., human resource accounting, the balanced scorecard or the intellectual capital (IC) approach. The IC approach is a general, holistic perspective to the intangible assets, i.e. the intellectual or knowledge capital, of a company. The approach is based on the observation that the market value of a company1 is usually higher than its monetary and non-monetary assets represented in traditional accounting. IC comprises immaterial values created by intellectual activities, e.g., human capital, customer capital, process capital and intellectual property. Some organizations, the best known probably being Skandia, have extended their reports on firm performance to include nonfinancial indicators, indicators of intellectual capital. Even though the IC approach provides a sound theoretical basis to determine the value of knowledge in organizations, the corresponding methods of measurement are (so far) pragmatic ones. The more abstract the notion of knowledge is, the harder it is to estimate its value. Still, the approach is used widely. Examples for concrete instruments to measure the IC of organizations are the Intangible Assets Monitor, the Intellectual Capital Navigator, the Skandia Navigator, the Balanced Scorecard, the Austrian “Wissensbilanz” (knowledge balance sheet) mandatory for e.g., Austrian universities, as well as single measures assessing intangible assets, such as Tobin’s q, the IC-index and the Calculated Intangible Value.

1.1.2 Definition Keeping the abundant classifications of knowledge in mind, it is clear that the conceptualizations influence the design of KM initiatives and the 1

The market value of a company is usually determined by the capitalization (value of the shares on the stock market) of a company.

Examples for IC instruments

16

1. Foundation

implementation of EKIs in many ways. Thus, we define knowledge broadly here and discuss some implications in detail: Definition of knowledge

Nature

Abstraction

Justification

Method

Relevance

Knowledge comprises (1-nature) all cognitive expectancies, no matter whether rational or used intentionally, (2-abstraction) i.e. observations about classes of phenomena (3-justification) meaningfully organized, substantiated and embedded in a context (4-method) through experience, inference or communication, (5-relevance) deemed valuable (6-agent) by an individual or organizational actor (7-application) to interpret situations and to generate activities. (1) Knowledge is a cognitive entity and refers to expectations about future events, acts or states. In organizational settings, knowledge used for interpretation and action not necessarily is true and, in the event of tacit knowledge, the knowing person might not even be aware of using it. Thus, when one takes the concept of knowledge from the scientific world to organizational settings, the rationality assumption that is pivotal in most, yet still debated philosophical positions on scientific knowledge is relaxed in order not to exclude tacit knowledge that is used unintentionally and thus lacks rational reflection. (2) Abstraction from individual situations, procedures, entities or generally phenomena is typical for modelling tasks that are at the core of the MIS perspective towards business organizations. The corresponding relationships between types and instances, entity types and entities as well as classes and objects point to the decision that although knowledge undoubtedly is bound to individual observations, knowledge focused in KM usually needs to be applicable for similar cases in order to be reusable. (3) Cognitive expectancies are contextualized information entities. Justification is based on (a) organization, i.e. structuring knowledge and linking it with other knowledge elements, (b) substantiation, i.e. enriching with evidence or (formal) proofs in order to increase its credibility, and (c) contextualization, i.e. considering the context of knowledge creation and application. This distinguishes it from information with respect to designing knowledge infrastructures compared to information infrastructures. (4) Justification is a non-trivial task that can be based on experience of individuals, logical inference or communication between individuals in order to achieve inter-subjectively justified beliefs. All of these processes can be supported by ICTs. (5) Not all thinkable knowledge is valuable in businesses and organizations. In a pragmatic view, KM concentrates on the portion of knowledge available in or accessible by organizations deemed valuable by its agents.

1.1. Knowledge

(6) Agent is meant here in the sense of an actor. Thus, both individuals or social entities such as teams or communities or entire organizations might act as knowledge-handling entities2. Examples of knowledge are scientific findings and theories, heuristics, rules of thumb, techniques, experiences, opinions, cultural customs and norms, world views. (7) Agents are always part of a social context which influences handling of knowledge of the actor and thus both, interpretation and actions. Put in a nutshell, knowledge can be defined as the capacity to interpret and act. Figure 1-6 summarizes our discussion and shows four central perspectives on knowledge in organizations, media to which knowledge is bound in these perspectives, a selection of seven paired types of knowledge which are used in knowledge processes and are supported by an EKI platform. In the following, types of knowledge and the medium to which knowledge is bound are discussed in detail. Types of knowledge. Figure 1-6 shows six types of knowledge which are discussed in the following. Source distinguishes between organization-internal and organizationexternal knowledge. Even though organizational boundaries are increasingly blurry, organization as a legal or social institution remains a focal point for the distinction between internal and external knowledge. Internal knowledge is knowledge that originates from within the organization either from its members or in the form of e.g., organizational routines or documented experiences. Organization-external knowledge is brought into the organization, either personally or in documented form. The accessibility dimension contrasts electronically accessible and electronically inaccessible knowledge. Knowledge published e.g., in a document management system can be accessed by all members of the organization that have access to this system whereas documented knowledge that is stored on the individual hard disc of a single employee cannot be found by interested knowledge seekers. Additionally, it refers to access to experts that hold knowledge about a specific domain. The security dimension comprises secured and unsecured knowledge. Higher visibility of knowledge, experts, networks and structures increases the risk that important knowledge is disseminated to competitors and threatens competitive advantages. Security refers to legal mechanisms such as patents and licenses, copyrights and trade secrets, organizational mechanisms such as incentives to employees, employee conduct rules and job design to secure knowledge and IT measures that prevent unauthorized access to EKI and prevent loss and manipulation of knowledge.

2

The term actor is preferred to agent as in the MIS literature agent regularly also refers to computer systems (intelligent agents). The old question whether computers can “think” and thus process and apply knowledge is out of the focus of this book (for a brilliant treatise of this topic see e.g., Dreyfus/Dreyfus 1986).

17 Agent

Application

Knowledge and EKIs

Source

Accessibility

Security

18

1. Foundation

Formality

The formality dimension ranges from formal, institutionalized, approved to informal, unapproved knowledge and reflects the degree of institutionalization of knowledge in an organization. Employees develop and apply knowledge independently of the formal approval system and might also share it within their community. This knowledge evolves as a group of employees commits to use knowledge in a specified way and is further institutionalized when knowledge is formally approved as part of the standard procedures in the organization. Business organizations rely on rules, roles and (standard operating) procedures, so there is a host of institutionalized knowledge which is applied by the organization’s members. The informal part of an organization’s knowledge base is rarely well supported and thus needs special treatment in EKIs.

is supported by

platform

Enterprise knowledge infrastructures

knowledge processes

identify

Knowledge formalize apply management organize refine life cycle distribute

informal, unapproved

(electronically) accessible secured formal, institutionalized, approved explicit

tacit specific, particular, contextualized object

Figure 1-6.

Knowledge individual

knowledge as product/ production factor

organizationinternal

expertise

abstract, general, decontextualized medium

(electronically) inaccessible unsecured

share

is used in

organizationexternal

types of knowledge

create

social system organizational resource, capability, core competence

Knowledge and its application in KM

intellectual capital

knowledge in organizations

feedback

1.1. Knowledge

19

Externalization turns tacit knowledge into explicit knowledge. Ever since Polanyi postulated that “we know more than we can tell” (Polanyi 1966), the tacit dimension has been popular. Many authors distinguish between tacit and explicit knowledge, however, Polanyi postulated that every knowledge has got a tacit dimension. Tacit knowledge is subconsciously understood and applied, difficult to articulate, developed from direct experience and handed on through conversation and shared experience (socialization, Nonaka 1991, 1994). Explicit knowledge can be formally articulated and shared through e.g., mathematical formulas, models or documents (combination). If explicit knowledge is documented, it is removed from its original context. EKI can help the receivers of explicit knowledge to reconstruct its context. Turning implicit into explicit knowledge is called externalization, turning explicit into implicit knowledge is called internalization. The distinction between tacit or sometimes called implicit and explicit knowledge helps to postulate different KM activities and different systems to support these activities. Figure 1-7 summarizes the processes of knowledge conversion.

Externalization

tacit knowledge

socialization tacit knowledge

from

explicit knowledge

Figure 1-7.

to

Knowledge conversion

explicit knowledge

externalization

share experiences: training on the job, brainstorming informal gathering

communicate intuitive, subjective experiences: metaphors, analogies, physical models

result: sympathized knowledge

result: conceptual knowledge

internalization

combination

avoid re-inventing wheel: handbooks, diagrams, learning by doing, stories, studies

share, integrate knowledge: text/ image/ audio/ videodocuments, discussion, formal training & education

result: operational knowledge

result: systemic knowledge

Processes of knowledge conversion (Nonaka/Takeuchi 1995)

The level of context of knowledge defines a continuum from specific, particular, contextualized knowledge describing one particular episode or event e.g., in a story to abstract knowledge, general, decontextualized knowledge captured e.g., in a mathematical formula. Before knowledge is distributed to a larger group of people, particular experiences can be generalized to lessons learned e.g., by extracting factors that might have influenced the outcome or by aggregating similar experiences to describe a practice (good or best practice). The more specific a knowledge element

Generalization

20

1. Foundation

is, the more context has to be provided by EKI in order for knowledge seekers to be able to understand, learn and reuse the knowledge element.

Knowledge as object

Medium. The medium on which knowledge resides can be an object, an individual or a social system. Knowledge as an object stresses the higher value of documented knowledge as opposed to data or (documented) information. Examples are lessons learned, best practices, experience databases, customized reports or context-enriched documents. In this perspective, knowledge is basically seen as information plus context, as networked information. The distinction between information and knowledge is a gradual one, a continuum. The common denominator of this perspective is that (a portion of the) knowledge used in organizations can be explicated and externalized and as a consequence untied from its creator and made available for “easy” reuse by other members of the organization.

Definition of knowledge element

A knowledge element is the smallest unit of (1) atomic, (2) explicit, (3) formally defined knowledge content.

Analysis of definition and examples

(1) A knowledge element is viewed as a single organized unit both from a conceptual and from a technical perspective. (2) It denotes a record of some form of externalization. (3) It is composed of a grouping of formatted information objects which cannot be separated without substantial loss of meaning together with meta-data describing the element. Examples for knowledge elements are concepts, categories and definitions (declarative knowledge), processes, actions and sequences of events (procedural knowledge), rationale for actions or conclusions (causal knowledge), circumstances and intentions of knowledge development and application (specific contextual knowledge). From an ICT perspective, examples are: a document, email message, instant message, video file, audio file, slide show or picture displaying an idea, proposal, recommendation, an expert’s opinion, a description of or solution to a specified problem3, a personal note with a write-up of a personal experience, a contribution to a forum, newsgroup, Wiki, Weblog or other form of CMS, an entry in a list of frequently asked questions (FAQs) and the answer to the question, an element in an experience database,

3

The stress is here on representing a solution to a specified problem. This is not necessarily a file, but can also be a selected portion, e.g., a document fragment, a video sequence or an audio theme.

1.2. Knowledge Work

a document with, e.g., a product presentation, lesson learned, good or best practice, story, study, write-up of an experiment, whitepaper, patent or report, e.g., about the results of a project milestone, a prototype, a model of e.g., a (business or knowledge) process, class, data, knowledge structure or other enterprise model, a learning object in a learning repository, a skill description in a skill database, an entry in a yellow page system or expertise locator describing available expertise on a specified topic, knowledge elements that connect some of the above elements to persons, groups, teams or organizational units, e.g., the description of skills of a particular employee or organizational unit, an evaluation of or a comment to one of these knowledge elements etc. Some authors mix the notion of knowledge as an object and explicit knowledge although explicit knowledge not necessarily has to be documented. Thus, we have to distinguish between the dimension relation to individual with knowledge either being part of an individual’s mind or separate as an object and the dimension explicitness with knowledge either being implicit and unreflected or explicit and thus communicable. Only explicit knowledge can be documented. Individual knowledge is nonobjectified knowledge bound to a single employee. The distinction between explicit and tacit knowledge stresses the fact that employees are not fully aware or cannot fully explicate what they know. Examples for categories of individual knowledge include skills, personal experiences and levels of expertise. In an organizational perspective, collective knowledge is viewed as knowledge bound to a social system. Organizational assets and resources other than individual knowledge are targeted by this approach. Also, the strategic process of selecting, fostering, exploring and exploiting strategically relevant organizational capabilities takes place on this level. Summing up, a well-founded concept of knowledge helps to study the possibilities to support the handling of knowledge processes by EKI. Therefore, the design and implementation of EKI depends on the KM initiative’s perspective on knowledge.

21

Individual knowledge

Organizational knowledge

1.2 Knowledge Work Already in 1997, an estimated 60% of the gross national product in the United States of America (US) was based on information as opposed to physical goods and services (Delphi 1997, 10). This is not surprising as the

Some data on knowledge work

22

Knowledge changes work

Case example as didactical instrument

Persona

1. Foundation

knowledge-intensive construction and development process of new products and services potentially determines 80 to 90% of the resulting production costs. There is a trend towards more complex problem-solving services in which the majority of employees are well-educated and creative, self-motivated people. Employees’ roles and their relationships to organizations change profoundly as so-called knowledge workers replace industrial workers as the largest group of the work force. 60% of US organizations think that between 60% and 100% of their employees are knowledge workers (Delphi 1997, 10). In 2008, about 80% of workers were employed in the service sector in the US (US Department of Labor 2008), about 72% in Germany (Federal Republic of Germany, Common Statistics Portal 2008) or 69% in Austria respectively (Statistik Austria). In the last decade, the highest absolute job growth in the US economy is noted in those professions labelled knowledge work (Wolff 2005). This scenario has been termed the information or knowledge economy and has dramatically changed valuation of knowledge work. The concept of knowledge work was coined in order to stress the corresponding changes in work processes, practices and places of employees and thus the differences to traditional (often manual) work. The following case example is taken from a field study conducted by one of the authors at a large German IT system house with 400 employees (Maier, Thalmann 2008). The case example at the end of this chapter (page 80) illustrates typical, although idealistic patterns of behavior with respect to how knowledge and learning is handled on the work place by knowledge workers. It is used in order to convey a rich picture of modern work places for graduates with a specialization in Information Systems and in what environment or work ecosystem EKIs are typically used. This form of description is called Persona and represents a concept mediating between real persons and roles. It is used in software engineering to aid discussions between customers and suppliers of software. Personas provide an instrument to keep in mind that knowledge workers are not all the same as in the traditional perspective on users that were all provided the same software solution, but require individual, also called personalized support that takes into account their distinct ways of handling knowledge or their individual knowledge routines. As a concrete example, a communicative, outgoing person might prefer direct communication to searching while a more isolated person might prefer first browsing through contributions in newsgroups. EKI design would consider both perspectives and attempt to provide bridging solutions that are favored by both Personas.

1.2.1 Definition and Characteristics Characteristics

Knowledge work is characterized as follows: target: solves ill-structured problems in complex domains with a high degree of variety and exceptions,

1.2. Knowledge Work

23

content: is creative work, requires creation, acquisition, application and distribution of knowledge and bases inputs and outputs primarily on data and information, mode of work: consists of a number of specific practices, such as expressing or extracting experiences, monitoring what can be learned from happenings, translating knowledge to other domains, interpreting and absorbing knowledge and networking with other people, personal skills and abilities: uses intellectual abilities and specialized knowledge rather than physical abilities and requires a high level of education, training and experiences resulting in skills and expertise, organization: is often organized decentrally using new organizational metaphors, such as communities of specialized knowledge workers, has strong communication, coordination and cooperation needs and is highly mobile, flexible and distributed, ICT: requires a strong yet flexible personalized support by ICTs. Knowledge work can be defined as follows: Knowledge work is an ideal type of work, i.e. an abstraction comprising key characteristics of a wide array of activities in organizations across occupations that creates, translates or applies new knowledge.

Definition of knowledge work

This definition is a rather narrow one so that only a small portion of work done in organizations would qualify as knowledge work. The broader term, information work, takes into account that not all work with information necessarily generates, translates or applies new knowledge and comprises knowledge work, management work and data (service) work (Drucker 1993, Schultze 2003). Data or service work relies on established procedures, is well defined and does not require equally high levels of education as knowledge work. Management work is performed by business owners, executives, legislators, senior officials and supervisors whose daily work practices comprise processing, communication and translation of (abundant) information and preparation, taking and execution of decisions. In this narrow view, knowledge work is restricted to (re-)producing new knowledge whereas data (service) work transforms information, but does not produce new knowledge. In a wider view, knowledge work comprises creation, acquisition, distribution and application of knowledge. Thus, it might be difficult to separate knowledge work from “mere” data (service) work in actual work practices so that EKI might be most useful when supporting information work in general, not “just” knowledge work. The term knowledge work refers to (Kelloway/Barling 2000): professions: occupations or job positions are classified into knowledge workers and non-knowledge workers or routine, manual etc. workers. This distinction is not without trouble because on the one hand all

Information work

Data/service work Management work

Knowledge work ...

... as professions, personal characteristics or activities

24

1. Foundation

human work requires some kind of knowledge and on the other hand even within one profession actual workers might differ widely according to the share of their work that qualifies as knowledge work. characteristics of a defined group of individuals: education, training and years of work experience are a necessity for a worker to be called an expert. In this case, knowledge work refers to experts’ work. However, on the one hand experts might not always be engaged in knowledge work and on the other hand less experienced employees might be engaged in just the same type of work than experts and therefore require just the same organizational and ICT design. activities/behavior in organizations: Thus, knowledge work should not be restricted to a certain class or group of employees. It should rather be used as a concept that allows a focus on commonalities across professions and positions for the application of KM instruments, KM-oriented organizational design and ICT support. Knowledge work consequently is regarded as an ideal type of work that serves as a model being well aware that concrete instances of knowledge work rarely will show all characteristics attributed to the ideal type. As a growing share of employees is increasingly engaged in instances of this type of work, the corresponding design of an ICT environment throughout an organization gains importance: an enterprise-wide infrastructure for knowledge work or, shortly, EKI.

1.2.2 Traditional Work versus Knowledge Work Environment for knowledge work

Organizational design

Many organizational theorists have approached the challenge of defining concepts for characterizing organizations in the post-industrial society by contrasting the traditional, i.e. industrial, work organization with new forms of work, in our case knowledge work. Table 1-3 compares the traditional, routine work environment of an office employee with the work environment of a knowledge worker. It shows changed requirements for organizational design and ICT support for knowledge work that have to be considered when designing EKIs and some aspects of economics that affect management of knowledge work. When compared to traditional work, knowledge work can be characterized by stronger communication needs, weakly structured and less foreseeable processes, assignment of multiple roles to one person rather than a single job position and increasing importance of teamwork in the form of project teams, networks and communities in addition to work groups and departments. Highly skilled (specialist) knowledge workers consequently envision their careers as sequences of projects (Barley/Kunda 2001) which they choose carefully according to personal preferences and opportunities to enhance their skills rather than a sequence of predefined steps within an organization’s hierarchy regardless of the activities that have to be per-

1.2. Knowledge Work

formed. Already in the early 90s, engineers in the US Silicon Valley on average switched employers every two years (Saxenian 1994). These changes are reflected by an organizational design that strengthens decentral units including a much more prominent role of projects. The boundaries of an organization are blurry and knowledge workers are engaged in a large number of communication, coordination and cooperation processes and practices that cross organizational boundaries. Alliances, joint ventures, (virtual) networks and professional communities are some examples for types of institutional settings that have been developed to organize these exchanges. Table 1-3. criterion

Traditional office work versus knowledge work traditional office work

knowledge work

organizational design orientation

data-oriented

communication-oriented

boundaries

organization-internal focus

focus across organizational boundaries, alliances, coopetition, (virtual) networks

structure

central, hierarchy

decentral, network, hypertext

process

highly structured, deterministic processes (pre-structured workflows)

ill-structured, less foreseeable processes (ad-hoc workflows)

group

work group, department

project team, network, community

role

one job position per person

multiple roles per person

career

sequence of job positions in a predefined, rigid hierarchy of a particular organization

sequence of projects including (frequent) switches between e.g., domains, employers, countries

type of contents

structured data (e.g., tables, quantitative data)

semi-structured data, e.g., links, hypertext documents, container, messaging/learning objects, workflows, skill directories

storage

(relational) database management system, data warehouse

document/content management systems, experience databases, newsgroups, mail folders etc.

ICT support

data handling coordination of accesses, integrity, control of redundancy modeling

synchronization, information sharing, distribution of messages, search and retrieval

data, business process, work- ontology, user profile, communiflow cation, activity/work practice

25

26

1. Foundation Table 1-3.

Traditional office work versus knowledge work

criterion

traditional office work

knowledge work

workspace

fixed workspace

mobile office (virtual office), multiple workspaces

equipment

personal desktop computer; poor resources

laptop, personal digital assistant, mobile phone; rich resources

connectivity

stand-alone

permanent, fast network connections, mobile devices

management focus

finance, past orientation, periodic reporting

balanced set, future orientation, instant access

location of value

things

flows

tangibility

tangible

intangible

metrics

production statistics, metrics for reporting

innovation statistics, metrics for managing

standardization

standards; standard products and services

common, yet customized products and services

economics

ICT support

Enriched modeling

From an ICT perspective, the main changes in the requirements occur due to the considerably higher complexity of data and the focus on organization-wide and inter-organizational communication and mobility of knowledge workers. Storage and handling of semi-structured data require additional ICT systems, e.g., document management systems or experience databases. Consequently, the challenges in handling data are no longer restricted to providing integrity, control of redundancy and coordination of accesses as in the relational database world. New challenges are complex synchronization needs of mobile workspaces, information sharing within and across organizational boundaries as well as search and retrieval in documents and messaging objects that are encoded in a large number of heterogeneous formats for semi-structured data and reside in a variety of data and document sources spread throughout the organization. Coordination in traditional office work is provided by workflow management systems that implement operative business processes. The lesser structured knowledge work can be coordinated by messaging and collaboration systems. Consequently, modeling used to focus largely on data (entity relationship modeling), objects and classes (object-oriented modeling) and business processes (business process modeling). Knowledge work requires content- and communication-oriented modeling techniques that define meta-data and provide taxonomies, ontologies, user models, communication diagrams, knowledge maps and diagrams that show what objects, persons, instruments, roles, communities, rules and outcomes are

1.2. Knowledge Work

involved in the main knowledge-related activities. Finally, the increased mobility of knowledge workers requires multiple, virtual workspaces that can be personalized according to the demands and practices of their users. This change in ICT support is backed by a corresponding major shift in the ICT infrastructure. Computers are no longer equipped with weak resources and used in an offline, stand-alone mode. Computers have rich resources, provide information-rich modes of interaction with users, permanent, fast network connections as well as highly flexible wireless and mobile connections and comprehensive communication features. Mobile devices, such as notebooks, PDAs, Blackberries and mobile phones are equipped with a wide range of applications. Summing up, this calls for (1) systematic, flexible handling of context, (2) intelligent analysis of vast amounts of substantially extended types of contents, i.e. semi-structured data in the organizational “knowledge base” and (3) extended collaboration. These functions have to be realized in or seamlessly integrated with the knowledge workers’ personal workspaces (section 5.1.2, 370ff). Correspondingly, management focus has shifted from a mere periodical financial focus with its past orientation to a flexible and balanced set of criteria that show the current status of the organization’s resources, processes, innovation and performance. The interest thus has shifted from tangible to intangible assets, from things to flows, from standards and standard products and services to common yet customized products and services (Skyrme 2000). Metrics are required not simply for reporting production statistics of goods and services, but for managing innovation processes. KM in this realm provides for more visibility of organizational resources, skills and knowledge processes and allows for a more systematic strategic management of (core) competencies in an organization. Consequently, KM initiatives primarily aim at fostering an organizational and ICT environment that is suited for knowledge work. Knowledge work is the primary target of KM, but corresponding organizational instruments and knowledge infrastructures might also aim at improving information work. This includes management and data (service) work. Substantially changed work practices of their largely increased main target group, knowledge workers, together with recent innovations in ICT infrastructure demand a strategic KM initiative. KM not only improves organizational effectiveness, but systematically realizes the potentials of a learning- or a knowledge-intensive organization for creating and sustaining superior competitive positions.

27

ICT infrastructure

Economics

Conclusion

28

1. Foundation

1.3 Knowledge Management The KM field has drawn insights, ideas, theories and metaphors from diverse disciplines. This section briefly reviews the history of KM resulting in a definition of the concept. Tracing the roots helps to understand the perspective which KM has or can have on organizations.

1.3.1 Roots of Knowledge Management KM as product of diverse disciplines

Organizational learning

Learning organization

Organization development

The roots of the term knowledge management can be traced back to the late 60s and early 70s in the Anglo-American literature. However, it almost took another 20 years until the term appeared again in the mid 80s in the context as it is still used today (e.g., Sveiby/Lloyd 1987, Wiig 1988). This time it got a tremendous amount of attention. The underlying concepts used and applied in KM, though, have been around for quite some time. There have been numerous fields and disciplines dealing with handling of knowledge, intelligence, innovation, change, learning or memory in organizations. Various approaches have played a role in the development of the theories of organizational learning, organizational memory and, ultimately, of KM. Basically, there are sociological, business and ICT lines of developing approaches for KM. The most important research fields that have contributed to KM are highlighted in the next paragraphs. Even though organizational learning (OL) has emerged as a field only in the 70s and 80s, it soon became a recognized way of looking at change processes in organizations. OL approaches share the hypothesis that phenomena of change in organizations are connected with collective or interpersonal processes of learning. OL definitions differ with respect to the question whether behavioral change is required for learning or whether new ways of thinking and new possibilities for action are sufficient. OL processes aim at weaving individual knowledge into organizational knowledge and can be classified into micro-organizational learning in groups and macro-organizational learning on the organizational level. Individual experiences and learning potentials are organizationally connected in groups (micro-structure). The groups’ learning results are then turned into organizational learning success (macro-structure). The term learning organization was coined to underline an organization’s skills in performing organizational learning, in more detail, its skills at creating, acquiring, and transferring knowledge, and at modifying its behavior to reflect new knowledge and insights. Organization development (OD) is a methodical strategy for intervention, initiated through consulting and planned by management assisted by a change agent concerning personal, inter-personal, structural, cultural and technological aspects. OL differs from traditional OD in that it considers change as the rule instead of the exception as in OD. OL views change as

1.3. Knowledge Management

endogenous, as part of the organization’s processes, and the (indirect) management of change is considered an organizational competence rather than an (external) expert’s competence. The basic idea of the organizational memory (OM) approach is that learning, whether individual or organizational, is not possible without memory. Memory, generally a system capable of storing things perceived, experienced or self-constructed beyond actual occurrence for later retrieval, is used as a metaphor here. OM is seen as a prerequisite for OL just as individual memory is a prerequisite for individual learning. As with many metaphors, the analogy between organizational and individual memory is weak. Thus, intuitive understanding of the term is often misleading, e.g., regarding OM as a “brain” to which organizations have access or the more technical analogy between computers and brains. OM simply means that employees, written records, or databases “contain” knowledge that is readily accessible. Emphasis has shifted from a static to an active interpretation of memory determining what an organization pays attention to, how it chooses to act, and what it chooses to remember from its experience. The results are individual and shared mental models as well as complex phenomena taking place when groups or organizations jointly “process” knowledge. Many approaches such as the transactive memory have been developed which claim to guide organizations to use their common or shared memory in a more efficient way. ICT is a key enabler for KM initiatives. Consequently, many researchers and practitioners in the field of AI have changed their research focus from expert and knowledge-based systems to KMS or EKI and, together with its psychological sibling cognitive sciences, try to establish the analogy between human and computer problem solving. Advanced AI technologies, such as formal logics, machine learning, neural networks, genetic algorithms and intelligent agents, provide “intelligent” KM tools, e.g., for semantic text analysis, text mining, user profiling and pattern matching. KM renews the old promise to provide concepts to improve the systematic handling of knowledge in organizations. KM can be viewed as a translation of OL and OM approaches to management terms and an integration with management concepts. It encourages goal-oriented design of interventions into the handling of knowledge, capabilities or competencies on a strategic, organization-wide level, redesign of knowledge-intensive business processes, improvement of personal KM as well as implementation of enterprise-wide ICT infrastructures for KM, i.e. EKIs.

1.3.2 From Data to Knowledge Management In addition to the interdisciplinary perspective on KM presented in section 1.3.1, another quite popular conceptualization compares KM to data and information (resource) management and postulates a development line between them which will be briefly surveyed in the following.

29

Organizational memory

Artificial intelligence

KM compared to OL and OM

30

1. Foundation

Many authors seem to agree on some form of a hierarchical relationship between data, information and knowledge (section 1.1, 3ff). Each higher level is based on or extends the preceding one. This conceptualization is used to postulate different demands for management, e.g., goals, approach, organizational roles, methods, instruments, and different resulting systems, e.g., databases, data warehouses, information and communication systems, KMS, on each of these levels. Historically, in the 70s and the beginning of the 80s the focus certainly was on data management (Figure 1-8). In the following, the steps will be discussed subsequently. knowledge organization

enterprise-wide horizontal data integration

technical data integration

isolation

Step 0

Step 1 use of DBMS

Step 4 information resource Step 3 conceptual information data integration separate responsi- management bility for data Step 2 • ‘90s data warehousing data data modeling/ data mining data standardi- management document • late ‘80s zation management data repositories reference models administration enterprise data enterprise resource • mid ‘80s modeling planning

data base administration isolated relational DMBS applications • m id ‘70s no special attention to data • beginning of IT

Figure 1-8.

Step 0: isolated applications

Step 1: database administration

information life cycle/ vertical data integration

very large DBS

OODBMS multidimensional DBMS active DBMS

Step 5 knowledge/ organizational memory knowledge management • late ‘90s/‘00s new ICT: KMS, CRM, portals application development with “intelligent” technologies content management Intranet/ICT platforms XML, RDF, OWL semi-structured data knowledge modeling DBMS and the Web XML DBMS content management systems

Historical development of information processing with focus on data (based on Ortner 1991)

The starting point for the historical development of information processing can be described by a joint consideration of program logic and data. There is no special attention being paid to data. Application systems hold their own data storages leading to redundancies and inconsistencies between different application systems. In the first step, technical issues therefore matter most. Database administration is concerned with the technical integration of previously isolated data storage units. Examples for tasks are to guarantee efficient data storage avoiding or controlling redundancies, to implement and administer database management systems (DBMSs) that provide instruments for technical integration between application systems or to tune the performance of database systems (DBSs).

1.3. Knowledge Management

As DBMSs penetrate organizations, semantic or conceptual data integration, data modeling and data handling are the most important questions to be resolved, called data administration. These tasks together provide semantic data integration which is the primary goal of step 2. Separate organizational units are established to coordinate data management tasks throughout an organization. In the 80s and 90s, this coincided with creating enterprise data models seen as instruments for an organization-wide integration of project or departmental data models. Information is understood as a production factor which has to be managed like other production factors (capital, labor). Thus, the scope of the information resource management is a much broader one compared to data management. The most important aspects are extension from managing syntactic and semantic to pragmatic aspects of information understood as an instrument for preparing decisions and actions, information logistics, the contingency approach to information, i.e. different interpretation of information in different situations, and the perspective-based approach to information, i.e. different user groups interpret the same data differently. An organization's ability to learn and handle knowledge processes (process view) or its ability to handle knowledge (product view) are considered the new key success factor. This requires new organizational design alternatives and also new ICT systems to support the smooth flow of knowledge which consequently have been called KMS. With the advent of an organization on a certain step, tasks introduced at a previous step still play a role. For example, database administration on step 1 covers not only hierarchical and network DBMS, but also relational DBMS (step 2), very large DBS (step 3), object-oriented, active and multidimensional DBMS (step 4) as well as XML DBMS, content management systems (CMS) and access DBMS from the Web (step 5, Figure 1-8). Knowledge management extends the focus of information management to the handling of new ICT, e.g., advanced database and AI technologies, as well as to enrich application development with intelligent technologies (Figure 1-8). With respect to data, KM needs to handle networks of semistructured, context-rich data, experts, participants and their combination. Meta-data and content management for semi-structured data on an enterprise-wide level need to be considered. Certainly, EKIs cannot be reduced to their data and meta-data structures, but offer a new variety of ways to support handling of knowledge in organizations. To sum up, in many organizational contexts and approaches in the literature, KM is viewed as the next consequent step in organizational information processing. Indeed, from a data-oriented perspective, this view can be justified. It explains, for instance, what data management tools and methods, what information logistics and ICT infrastructures are required to effectively build EKIs. A KM definition needs to consider this view and the broader OL view that includes organizational functions and processes traditionally not part of information management (section 1.3.1, 28ff).

31 Step 2: data administration

Step 3: data management

Step 4: information management

Step 5: knowledge management

Extended tasks on lower steps

Extensions by KM

KM next step in information processing

32

1. Foundation

1.3.3 Definition KM is still a young field with multidisciplinary roots. Thus, it is not surprising that there seem to be almost as many definitions to the term than there are approaches or “schools” of authors contributing to the field. Basically, there are two groups of KM approaches, human- and technologyoriented KM approaches which basically reflect the origin of the approaches, either in a human/process-oriented OL, organization science background, or in a structural organization science, MIS, computer science or AI background. There is also agreement that there are more holistic KM conceptualizations which encompass both directions. Figure 1-9 shows the two sides of KM and some examples for concepts developed in holistic approaches aimed at their integration. strategy organizational knowledge human-oriented knowledge management

individual knowledge

KM tools knowledge life cycle business and knowledge processes

technology-oriented knowledge management

platforms

integrating instruments

Figure 1-9.

Human versus technology-oriented KM and approaches to their integration

The term management is used here in a functional sense (managerial functions approach) in order to describe what needs to be done, e.g., planning, organizing, leading and controlling as opposed to an institutional sense (managerial roles approach) which describes who is responsible for management tasks and roles. This leads to the following definition: Definition of knowledge management

Knowledge management is defined as the management function responsible for regular (1) selection, implementation and evaluation of knowledge strategies (2) that aim at creating an environment to support knowledge work (3) internal and external to the organization (4) in order to improve organizational performance. Implementing knowledge strategies comprises (5) person- or organization-, product- or process-oriented instruments (6) suitable to improve the organization-wide level of competencies and ability to learn.

Strategy content

(1) Systematic interventions into an organization’s knowledge base have to be tied to business strategy. Knowledge strategies guide the imple-

1.4. Knowledge Management Instruments

mentation of a KM initiative and tie it to business strategy. (2) KM creates an infrastructure, organizational and technological, to improve knowledge work. EKIs are part of this infrastructure. (3) Knowledge processes are not restricted to the organization’s boundaries, but involve cooperation with partners, suppliers and customers. (4) KM mainly aims at improving organizational effectiveness. However, creating, maintaining or distributing intellectual capital results in a higher valuation of an organization. (5) Depending on the perspective on KM, objects of the implementation of knowledge strategies can be all media of knowledge (section 1.1.2), i.e. objectified knowledge resources (documented knowledge), people and organizational or social structures. The environment of knowledge work that is to be created or fostered by KM and thus the instrument that is used for the intervention can be distinguished into the level of individuals (person) or collectives (organization) as well as into knowledge as a product or as a process (section 1.4). (6) KM is not exclusively about individual learning. Collective learning is (a) of differing types, e.g., single loop, double loop, deutero learning (Argyris, Schön 1978), (b) takes place on different levels, e.g., work group, project, community, organization, network of organizations, and (c) in different phases, e.g., identification, creation, diffusion, integration, application, feedback. KM aims at the organizational competence base and the ability to continuously improve it. None of these areas explicitly focuses on the contents, that is the actual subjects, topics, domains or knowledge areas for which a KM initiative and finally an EKI builds a supportive environment. The reason for this is that the definition of KM should support all kinds of knowledge areas. Summing up, there are substantial benefits to be gained from a KM initiative. An implementation of ICTs to support a strategically relevant KM initiative not only has to select a KM perspective and often a combination of KM tools and systems, but also integrate KM instruments with the supporting technology.

1.4 Knowledge Management Instruments As explained in the definition of KM in section 1.3.3, implementation of knowledge strategies requires systematic interventions with the help of instruments, called KM instruments. Section 1.4.1 gives a definition of this term. Sections 1.4.2 and 1.4.3 describe some central product- and process-oriented KM instruments, structured according to the classification introduced in section 1.4.1. Section 1.4.4 finally presents the model of knowledge maturing that is used later on to structure the central layer of knowledge services in the EKI architecture that will be studied in detail in chapter 4.

33

Strategy implementation

34

1. Foundation

1.4.1 Definition

Human-oriented perspective

Technologyoriented perspective

Definition of knowledge management instrument

Analysis of the definition

Even though the terms KM instrument, KM project, KM initiative and KM measure are widely used, there is hardly any concrete definition of any of these terms. A large number of measures has been proposed as part of case studies on KM which also comprise more traditional person-oriented measures well-known in human resource management, e.g., programs for personnel development, content-oriented measures well-known in database theory that revolve around using (simple) meta-data, organizational measures well-known in organization science, e.g., job rotation, job enrichment or ICT measures well-known in MIS, e.g., using databases, email or collaboration tools. Several case studies deal with the introduction of KM in organizations and describe what instruments were used. KM instruments target different goals and consist of several measures that have to be aligned and supplement each other. Instruments often comprise organizational as well as technological measures. Thus, it is useful to review a human- and a technology-oriented perspective on KM instruments before aiming at a comprehensive definition of KM instrument. Instruments for knowledge organization are intervention tools that are describable, get deployed purposefully, traceable for an observer, have a clear knowledge orientation and are relatively independent of the organized knowledge (Roehl 2000). Implementation of knowledge strategies is seen as a purposeful intervention into the way an organization handles knowledge. Clear knowledge orientation distinguishes KM instruments from other tools. Knowledge refers to contextualized information in an ICT context. KM instruments have to provide context in order to show knowledge orientation. Finally, a KM instrument in this view has to be general, spanning knowledge domains rather than being domain-specific. KM tools are based on technologies, broadly defined, which enhance and enable tasks along the knowledge life-cycle. As with any tools, they should be designed to ease the burden of work and to allow resources to be applied efficiently to those tasks for which they are most suited. It is important to note that not all knowledge tools are computer-based. Pulling these two perspectives together leads to the following definition: A KM instrument is (1) a collection of organizational, human resources and ICT measures that are aligned, (2) clearly defined, (3) can be deployed purposefully in an intervention into an organizational knowledge base in order to achieve knowledge-related goals, (4) target contextualized information as object of intervention and (5) are independent of a particular knowledge domain. (1) Only parts of the valuable knowledge assets exist in explicit form as documented, electronically accessible knowledge. Therefore, KM instru-

1.4. Knowledge Management Instruments

ments have to consider person-oriented measures. Organizational measures are implemented, e.g., as rules, roles, procedures and newly or redefined processes that describe how to deal with ICT systems. Last, but not least, the focus here is on those KM instruments that are enabled, fostered or substantially supported by ICTs. (2) Clearly defined means that any proposed instrument has to clarify what measures and tools are involved so that it is possible to decide if an observed phenomenon in an organization matches this definition. (3) KM instruments have to be applied with clear knowledge goals in mind, usually within the frame of a systematic intervention with the help of a KM initiative. Organizational knowledge base reflects people’s skills, contents, processes and (ICT) tools and systems in an organization that support handling of knowledge. (4) Knowledge orientation of the KM instrument can only be accomplished if contents of ICT systems are “knowledge-prone”, thus being contextualized information instead of only data. An example is a database containing experiences, lessons learned or best practices together with links to people who made these experiences and/or experts in the domains that are described (knowledge) as opposed to a database holding telephone numbers of employees (data). Embedding information into context is crucial. In ICT systems, it can be achieved by assigning appropriate meta-data and systematic management of an ontology to help users to integrate information into their personal knowledge bases (chapter 3). (5) Finally, a KM instrument should be independent of a specific knowledge domain and can be targeted at any topic or (core) competence of an organization. Figure 1-10 gives an overview of KM instruments that have been proposed in the literature and are applied widely. KM instruments usually target either individuals (person) or collectives (organization) along the dimension organizational level and knowledge as object, in the form of a product or knowledge in a process-oriented, encultured form, i.e. practices, processes or routines. All examples of KM instruments are supported by ICTs. Person-oriented KM instruments primarily aim at knowledge that is provided by, managed by or bound to individuals, e.g., personal experiences or routines, ideas, proposals, self-managed ad-hoc learning processes or meta-knowledge about individual skills. Organizational KM instruments target knowledge that is jointly created, shared, integrated, validated, legitimated or committed and thus is bound to social systems. Social systems in organizations are described with formal organization design, especially business and knowledge processes supported by good or best practices, knowledge maps, knowledge process reengineering and process warehouses, projects and work groups supported by lessons learned and case debriefings as well as informal organizational design, reflected by communities and knowledge networks supported by the instrument communities.

35

Dimensions

Person

Organization

36

1. Foundation

Semantic content management provides the infrastructure for knowledge processes whereas learning processes are systematically supported by technology-enhanced learning. expert advice process (knowledge in routines)

personal knowledge routines self-managed ad-hoc learning competence management

product (knowledge as object)

knowledge process good/best reengineering practices technologyenhanced communities learning case debriefings

knowledge maps

idea & proposal management experience management person (knowledge bound to individuals)

lessons learned

semantic content management organization (knowledge in social systems)

Figure 1-10. Knowledge management instruments

EKIs aim in general at providing a platform for KM and in particular foster implementing knowledge strategies with the help of a defined set of KM instruments. In the following, the identified KM instruments are structured into those that target knowledge as a product (section 1.4.2) versus those that target knowledge as a process (section 1.4.3).

1.4.2 Product-oriented Instruments Documented knowledge certainly is of high importance with respect to EKI design. On the one hand, product-oriented KM instruments target personal knowledge, such as personal experiences, ideas, proposals and skill descriptions. On the other hand, documented knowledge can be spread across multiple sources and requires identification and visualization with the help of knowledge maps as well as integration supported by ontologies. Ontologies also aid managing semantic content. While this instrument targets electronically available content as potential knowledge sources throughout the organization, there are two instruments that specifically establish the systematic handling of inter-subjective knowledge with commitment, i.e. case debriefings and lessons learned.

1.4. Knowledge Management Instruments

37

Implementation of experience management systems eases documentation, sharing and application of personal experiences in organizations. These systems have to be integrated into daily work practices of employees in order to be accepted. Several approaches exist that support capturing of experiences, e.g., information mapping, learning histories or microarticles. On an organizational level, systematic management of personal experiences enables a company to solve recurring problems more effectively. However, there are barriers preventing documentation or reuse of experiences. Time required for documenting experiences is critical because it imposes additional efforts on employees. Therefore, organizational measures are required that provide time tolerances and keep the effort as low as possible. Simultaneously, sufficient context of the experience has to be provided. ICT solutions help to automatically detect context. Personal barriers, e.g., insufficient willingness to share knowledge or to apply knowledge created by other employees (not-invented-here-syndrome) have to be considered by measures like trust management and incentive systems. Organizational proposal systems are formally defined processes that collect and handle ideas and proposals for improvements submitted by employees. Experts review proposals and evaluate them in a committee. If the proposal is selected, the employee is rewarded, mostly financially, and it might then be implemented. A template can help to structure proposals and an automated workflow can identify appropriate experts for reviewing them. From an ICT perspective, a database system as a minimal solution can be used to store proposals. Semantic content management can help interpret the proposals, e.g., with a glossary for acronyms and special terms probably not known by reviewers of different areas of expertise. Competence management supports systematic analysis, visualization, evaluation, improvement and usage of competencies held by individuals. Competence management comprises expertise locators, yellow and blue pages as well as skill management systems, also called people-finder systems. Skill management helps to make skill profiles accessible, identify learning paths for employees and ensure their currency and relevance. A central skill ontology, also called skill tree, has to be defined that provides context for all existing, required and wanted skills in the organization. Training measures have to be offered. Skill management systems are often not limited to information about skills, their holders and their skill levels, but also contain information about job positions, projects and training measures in which employees learned, used and improved their skills. Yellow and blue pages are directories of organization-internal and -external experts. Profiles of experts together with contact details are listed according to knowledge domains for which they may be approached. Information about employees’ skill levels and degrees of expertise can be used, e.g., to connect people, staff projects, filter and personalize EKI contents and services.

Personal experience management

Idea and proposal management

Competence management

38

1. Foundation

Semantic content management

Semantic content management refers to managing meaningfully organized content, i.e. documented knowledge embedded in a context. The term semantic means that content is well-described with the help of metadata that assigns meaning and structure to the content and that these descriptions can be interpreted by machines and used for inferencing. The instrument is tightly related to an IT solution, but there have to be rules that guide definition and use of semantics, monitoring external knowledge sources for interesting content, developing an appropriate content structure as well as publishing semantically enriched documents. Semantic content management also allows for “smart” searching, collaborative filtering and can be integrated with competence management in order to handle interests to connect people by a joint analysis of semantic content and skills. Knowledge maps can be used to aid access to knowledge, to knowledge sources or to knowledgeable persons. Central goal in this instrument is creation of corporate knowledge directories which visualize and support more efficient access to and handling of knowledge. Main objects of mapping are experts, project teams, networks, white papers or articles, patents, lessons learned, meeting protocols or generally document stores. Knowledge source maps visualize locations of knowledge, either people or systems and their relation to domains or topics. They can be further classified into knowledge topographies to identify gaps, competence maps to find experts and pointer systems that directly link from challenges within a process to a contact that can assist. Knowledge asset maps visualize amount and complexity of knowledge that a person or system holds. Knowledge structure maps show relationships between knowledge domains and should not only visualize that there is a relationship, but also explain its type. Formal definition of knowledge structures results in ontologies and supports integrating diverse knowledge sources. Knowledge mapping can also be used in order to highlight knowledge processes, especially processes of knowledge development and application. These maps are combinations of process models and knowledge source maps. Knowledge development maps visualize processes or learning paths that can or have to be performed by individuals or teams in order to acquire skills. Knowledge application maps describe what process steps have to be performed in what situation at what step in a business process, e.g., who should be contacted for a second opinion. Lessons learned are the essence of experiences jointly made and systematically documented, e.g., in projects or learning experiments. In a process of self-reflection, such as at the end of a project milestone, called after-action reviews, or at the end of a project, called project debriefings, project members jointly review and document critical experiences made in this project. Lessons learned can also aid individual self-reflection about one’s own experiences, but primarily aim at joint reflection that explicates know-how gathered in a team and learning from the experiences of others.

Knowledge maps

Lessons learned

1.4. Knowledge Management Instruments

Lessons learned are thus the product of a formal process that involves a collective of project members who share, discuss, reflect, verify as well as integrate their experiences and finally commit to them. This process can be moderated by a lessons learned coach. Templates can be created that support structured documentation and help the team to include important context. A system can aid this process and store and provide access to all documents containing lessons learned. A subject matter expert could review the documents and further enhance them by referencing other documents, projects and people. Rules support integration of the lessons learned instrument into standard project plans and can also enforce that project managers study lessons learned before starting a new project. Whereas lessons learned aim at systematically eliciting experiences made by teams in projects, case debriefings target experiences documented by work groups in business processes. Generally, the term case can be applied to a wide variety of phenomena about which knowledge is documented. From a business process-oriented perspective, a case is an instance of a business process with an explicit connection to a customer. Thus, this instrument focuses knowledge gained in specific, interesting cases encountered during operative work in business processes. In extension to business process definitions that abstract from the specifics of individual cases, case-oriented knowledge can enrich a process warehouse. As the knowledge is assigned to specific business processes, templates and rules can be developed that structure the types of cases that can be encountered and help to document case knowledge. Coaches can help employees to document cases and the experiences can be reflected in the work group responsible for the business process (commitment by work group) or by process managers (legitimation by supervisor). From an ICT perspective, several systems, particularly a case database system and, in formally structured environments, case-based reasoning systems aid retaining, searching and retrieving case knowledge.

1.4.3 Process-oriented Instruments Whereas product-oriented KM instruments target different types of documented knowledge in the sense of objects that can be accessed and reused not unlike information objects, another group of KM instruments aims at knowledge in a process-oriented form. This includes (1) retaining knowledge in a process-oriented form, e.g., personal knowledge routines, good or best practices, (2) directly targeting the design of knowledge and learning processes, e.g., expert advice, knowledge process reengineering or technology-enhanced learning or (3) informal organizational routines that aim at improving individual learning, e.g., self-managed ad-hoc learning or sharing of knowledge in communities. Even though some of these instruments also involve knowledge in an objectified form, e.g., communi-

39

Case debriefings

40

Personal knowledge routines

Self-managed ad-hoc learning

Expert advice

1. Foundation

ties might have a community home space, the primary focus is on supporting processes of handling knowledge, rather than documenting knowledge in a content or container fashion. Even in knowledge work, certain knowledge-oriented activities follow typical patterns. Knowledge routines comprise existing, allowed, recommended or prescribed partly routinized activities of knowledge work. Routines can be structured and made available for reuse, e.g., by knowledge brokers. Bundles of services might partly support knowledge routines for expressing, supported by templates and contextualization measures, translating, acquiring internal and external knowledge, integration and validation measures for knowledge of diverse sources, monitoring, getting an update on and awareness for current activities in an organization with respect to a process, a project or a domain and networking, supported by collaboration technologies and by competence management. The ICT infrastructure can support individual reuse of knowledge routines even though they are personal in the sense that employees individually handle their own routines. Organizational instruments can also aim at managing the transition process from personal knowledge routines to team, work group or unit good practices. This KM instrument reflects a specific type of personal knowledge routine stressed here because of the supposed tremendously increasing importance of individual, ad-hoc, self-managed learning processes, particularly the ones on the job, directly at the workplace. The instrument can provide support for personal learning processes, e.g., with the help of structuring and offering learning objects, learning paths and reflecting on learning activities by peers and experts within the organization or even crossing its boundaries. It can thus be part of comprehensive technology-enhanced learning instruments that are implemented in an organization. Expertise is often readily available, particularly in larger organizations, but meta-knowledge about who knows what is the bottleneck for an efficient and timely solution to knowledge problems. The instrument establishes a formal procedure that enables employees to pose requests for knowledge. A template structures questions and ICTs, e.g., a forum, can provide support for quick accessibility to unanswered questions. Semantic content management might be used to scan open questions and draw the attention of appropriate experts to the questions. Standard operating procedures for expert advice might differentiate between ordinary requests answered as soon as possible and urgent requests for which handling is guaranteed within an agreed time frame. Responses are given by whoever believes to have a solution to the posed problem. In case of urgent requests and if no response is submitted within a certain time frame, the question is relayed to an identified (subject matter) expert. The instrument requires

1.4. Knowledge Management Instruments

primarily organizational measures, but can also be supported by a forum or other content management system. Supporting or enhancing learning through ICTs has a long tradition (section 4.5, 331ff). E-learning, a sibling of e-business or e-government, is learning supported by multimedia or hypermedia contents that are online accessible backed by functions enabling communication between relevant user groups. Technology-enhanced learning is a more recent term that emphasizes that learning is not automatized, but that learning processes are supported and fostered by ICT. Newer approaches stress the importance of reusable learning material in the form of learning objects, the role of collaborative technology in interactive learning processes between and among teachers, tutors, coaches and learners, adaptive, adaptable and personalizable learning solutions as well as situation-oriented deployment of learning material in on-demand, workplace or ambient learning solutions. The instrument is traditionally not targeted as a KM instrument due to the fact that despite numerous attempts to bridge the gap between the two intuitively strongly related fields of e-learning and KM, they are still quite separated in research and practice. In large organizations, e-learning and KM are institutionalized in different organizational units, systems as well as attitudes towards handling knowledge. A more formal, elaborate and resource-intensive training approach with pre-defined courses contrasts a less formal, leaner approach, e.g., “harvesting” knowledge in projects and directly handing it on to an unspecified target group without much effort put into validating, didactically refining or examining success of the learning processes. Due to the fact that KM and e-learning are both approaches that intend to improve construction, preservation, integration, transfer and (re-) use of knowledge and competencies, the latter is integrated here as a KM instrument. However, a variety of approaches within e-learning might also be considered as individual KM instruments in their own right, e.g., developing courses with certification, peer or informal learning or selfmanaged, ad-hoc learning. Lessons learned collect project experiences in a descriptive, rather than a procedural way. The sharing of (good or) best practice is an approach to capture, create and share experiences or know-how proven to be effective which supposedly can be applied elsewhere in a process-oriented form as e.g., procedures, task patterns or workflows. As managers might argue about what exactly is “best” in a practice, several organizations use different levels of best practice, e.g., (1) good (unproven) idea, (2) good practice, (3) local, (4) company, (5) industry and (6) world best practice. These categories reflect the scope in which the corresponding practice has been deemed valuable or has been selected as best in a bunch of candidate practices. So-called best practice teams are permanent institutions within an organization’s networking infrastructure. They provide guidelines about what constitutes good or best practices and support identification, transfer,

41

Technologyenhanced learning

Good/best practices

42

Communities

Knowledge process reengineering

1. Foundation

implementation, evaluation and improvement of practices. Goal is continuous process improvement, thus best practices may lead to redesigned standard operating procedures, core and support business processes and knowledge processes. Community management targets creation and fostering of communities or knowledge networks. Communities differ from knowledge networks with respect to who initiated their foundation. Communities are founded by like-minded people (bottom-up) and can at most be fostered by superiors. Knowledge networks are established and legitimated by management (top-down). However, organizational and ICT measures to foster communities are the same as the ones used to support knowledge networks. Communities per definition cannot be controlled or externally induced. Instead, organizations can provide employees with time and space to share thoughts, establish IT tools, e.g., community home spaces, blackboards, Wikis or other forms of specifically designed content management systems that support exchange of thoughts and create new roles like community managers that help keeping discussions going and look for important topics that should gain management attention. Knowledge process reengineering (KPR) aims at redesigning business processes from a knowledge perspective. The term references business process reengineering (BPR) that aims at fundamental (process innovation) or evolutionary (process redesign) changes of business processes with the goal to increase organizational effectiveness. In addition to traditional BPR instruments, knowledge-intensive business processes are partially improved by KPR. The focus is on designing knowledge processes that connect business processes, defining cooperation scenarios, improving communication patterns between employees, as well as on “soft” skills or an organizational culture supportive of knowledge sharing. Business process models are stored in model bases. The model base can be expanded so that it handles not only knowledge about the process, but also knowledge created and applied in the process. This is termed process warehouse which can be used as a foundation for KPR. Examples for contents in process warehouses are exceptional cases, case-based experiences, reasons for decisions, checklists, hints, frequently asked questions and answers, potential cooperation partners or suggestions for improvements.

1.4.4 Knowledge Maturing From an ICT perspective, numerous systems aim at improving knowledge and learning processes as well as organizational competency development which are typically designed and managed according to the specific needs of the respective organizational units. Employees thus use a fragmented systems landscape in which each system supports a certain part of knowledge and learning processes. There are substantial conceptual challenges of designing learning and knowledge processes that bring together the sep-

1.4. Knowledge Management Instruments

arated organizational support infrastructures fostered by different organizational units. Organizational units and corresponding application systems typically target knowledge of different degrees of maturity. Pruning the tree of types of knowledge elements and guiding employees on how to use the channels of knowledge transfer is thus a pivotal task in any KM initiative. In the following, the knowledge maturing process is described in order to provide a framework for designing integrating types of knowledge elements, processes and channels in KM. In a first step of structuring this process, Figure 1-11 shows the five phases that have been identified after analyzing some practical cases as well as on the basis of a large empirical study (Schmidt 2005, Maier 2007, Maier, Schmidt 2007). drives acquire motives curiosity, creativity actions explore, experiment practices monitoring

steps

investigation

knowledge map knowledge elements

bond defend self-realization social belonging appropriate, commit endorse, validate, share networking personalization

idea news

individuation

interaction

personal knowledge routine

community artifact

personal experience

expert advice

43

Knowledge maturing process

pass-on comprehend status power arrange, learn, certify formalize, approve translating expressing

instruction

in-form-ation

case debriefing

learning object

lesson learnt

course

good / best practice

test

Figure 1-11. Knowledge maturing model

Figure 1-11 lists a number of key concepts that further explain the individual steps of the knowledge maturing model. The steps are further differentiated with the help of: Evolution theory postulates that during evolution humans have developed a set of innate psychological mechanisms that drive their behavior: acquire, defend, bond, comprehend (after Watson 2008) and pass-on. There are numerous motivation theories explaining human behavior as aim to satisfy a set of motives. Motives that drive the knowledge maturing process are social belonging, power, status, curiosity and self-realization. The steps typically involve many (knowledge) actions, e.g., access, collect, converge, coordinate, create, discuss, distribute, evaluate, identify, inquire, network, prepare, request and review (Hädrich 2008). However, there is a set of distinguished actions which describe the core of the step, i.e. the main activities that persons engage in when pursuing a certain step. Finally, the steps can be characterized with the help of the primarily targeted informing practice: expressing, monitoring, translating, networking (Schultze 2000) and personalization, i.e. marking a knowledge element as

Drives

Motives

Actions

Informing practices

44

Knowledge elements

Investigation

Individuation

Interaction

In-form-ation

Instruction

1. Foundation

one’s own so that one can trace back future developments towards an individual’s creation in order to have the individual benefit from it. Organizations typically handle large numbers of knowledge elements for a variety of reasons. Some of them can be classified into types of knowledge elements that are typical for individual steps of the knowledge maturing process. These knowledge elements provide a direct link to the KM instruments described in sections 1.4.2-1.4.3. Phases. The individual phases are described in the following. New ideas are developed by individuals either in highly informal discussions or by browsing the knowledge spaces available inside the organization and beyond, particularly in the World Wide Web. This step is driven by curiosity and creativity. The knowledge is subjective and deeply embedded in the context of the originator. The vocabulary used for communication is vague and often restricted to the person expressing the idea. This step, although considered important, has been neglected in academic discussions as well as in many practical implementations of EKIs. New ideas or results found in the investigation phase that have been enriched, refined or otherwise contextualized with respect to their use are now appropriated by the individual. This means that the individual marks its contributions so that it can benefit from its future (re-)use. The experience made is thus personalized. This step is driven by social motives and the benefits that individuals typically attribute to sharing of knowledge. These are, among others, belonging to a preferred social group, thus increasing probability of getting back knowledge from the community when one needs it. From the perspective of semantics, this accomplishes an important maturing step, i.e. the development of common terminology shared among community members, e.g., in discussion forum entries or Blog postings. Artifacts created in the preceding two steps are often inherently unstructured and still highly subjective and embedded in the context of the community. In this phase, purpose-driven structured documents are created, e.g., project reports or design documents or, with a stronger knowledge connotation, rich case descriptions, lessons learnt or good practices, in which knowledge is desubjectified and the context is made explicit. Documents produced in the preceding step are typically not well suited as learning materials because no didactical considerations were taken into account. Now the topic is refined to improve comprehensibility in order to ease its consumption or re-use. The material is ideally prepared in a pedagogically sound way, enabling broader dissemination. Individual learning objects are arranged to cover a broader subject area and thus are composed into courses. As a consequence, this subject area becomes teachable to novices. Tests and certificates confirm that participants of formal trainings have achieved a certain degree of proficiency.

1.5. Information System Architectures

45

Knowledge thus can be classified according to its level of maturity. The class then suggests the appropriate form of learning and technical support. Classes of knowledge. The following criteria have been identified as useful to define classes of knowledge: Certainly, the most obvious categorization refers to a validation process and could distinguish in a first step between unproven and proven knowledge. In a more refined version that considers the specifics of organizational knowledge, validation could take into account the number of successful uses of knowledge, systematic tests or, (mathematical) proves. In analogy to mineralogy, this criterion describes the (alleged) reliability of information or knowledge. According to Watson (1999), a possible scale runs from unidentified sources of rumors up to stock exchange data. With deepened understanding, connections to other topics become visible. This must not be confused with inherent contextualization of knowledge which decreases in the knowledge maturing process and refers to the degree of implicit linkage to the creation context, so that it cannot be used outside the original context. Inherent contextualization and inter-connectedness are inverse properties. Knowledge can be structured according to the amount of support it gets. Support can be in the form of commitment by members of groups, teams, communities or other organizational units. Another form of support can be authorization to use knowledge by supervisors, executives or committees as well as legalization and standardization, forms of legitimation. As knowledge maturing is basically interconnecting individual learning processes in which knowledge is taught and learnt, an important criterion is teachability. Whereas immature knowledge is hard to teach, even to experts, formal training allows by definition for wide-range dissemination.

Validity

Hardness

Context

Commitment/legitimation

Form of learning

1.5 Information System Architectures Architectures play an important role in the discipline Management Information Systems as blueprints or reference models for corresponding implementations of the organization-wide IS landscape. The term architecture as used in MIS origins in the scientific discipline Architecture. An architecture represents the structure of a system by means of its main components and the essential relationships between them. Table 1-4 gives examples for types of architectures used in organizations to design or document ISs. Architectures are the basis for communication about a system between stakeholders during the design process, e.g., between members of a steering committee deciding on essential interfaces and interactions between several ISs or IS projects. Architectures document fundamental and often irreversible design decisions. They can be used as a plan or draft for sys-

Types of architectures

Plan and documentation function

46

1. Foundation

tems and thus prescribe what these systems should look like (plan function). Architectures can also describe (parts of) already existent ISs, e.g., the main components or modules of a standard software package together with potential interactions with systems provided by other vendors (documentation function). . Table 1-4.

Types of architectures

type

description

application

functions, methods, modules, programs or entire application systems with relations between them

data

entity types and relationship types as an (ideally enterprise-wide) abstraction of multiple data models describing individual database systems

information system

components, e.g., application, communication, data, technology, of an enterprisewide information infrastructure and rules that help to implement IS strategy

network

computers, routing devices and network media connecting them

technology

hardware, software and network architecture showing how to implement the logical architectures with information and communication technologies

Conceptual architectures

Physical architectures

Architectures for enterprise knowledge infrastructures

Section overview

Architectures can be classified into conceptual and physical architectures. Conceptual architectures guide the implementation of physical architectures. Examples for conceptual architectures are application, data or IS architectures. The latter describes the entire information infrastructure of an organization, also called information landscape, from a logical perspective limited to the most important components, relationships and characteristics of the system. Examples for physical architectures are hardware, software or network architectures. These architectures describe the main components that are actually implemented in an organization together with the relationships between them. Physical architectures have been important instruments for planning an organization’s computing facilities, in particular computer servers, personal computers and the network connections between them. So far, conceptual architectures primarily aim at a documentation or a plan of the entire information infrastructure supporting business processes in organizations, e.g., the architecture of integrated ISs consisting of goal, data, function, organization, control and output perspectives (Scheer 2001). EKI architectures are seen as complements to transaction-oriented architectures. Both types of architectures together describe or prescribe an enterprise-wide information and communication landscape. The construction of a separate EKI architecture sheds light on the currently underdeveloped handling of semi-structured data, communication, cooperation and learning required in knowledge work. Goal of this section is to discuss some architectural alternatives and related concepts important to understand the EKI architecture presented in

1.5. Information System Architectures

47

section 1.6. Section 1.5.1 presents centralistic, distributed or peer-to-peer as well as agent architectures. Section 1.5.2 discusses approaches to classify systems in an enterprise architecture, namely the IS pyramid, the value chain and an example for an integrated, service-oriented architecture (SOA) of an enterprise resource planning (ERP) system.

1.5.1 System Architectures System architectures represent the structure of software applications on an abstract level. Components are delimited parts of software systems, e.g., functions, procedures, abstract data objects, abstract data types or object classes. There are numerous views on what constitutes a system or software architecture, particularly what are major components, how does the structure look like and which decisions are hard to change or irreversible. Four major views on software architecture can be distinguished (Clemens/Northorp 1996): the conceptual view represents the user's view and functional requirements and tasks of the system, the module view depicts the structure of software modules for software developers, the process view describes control and communication flows and allows to analyze performance, throughput and synchronization of processes or functional groups, the physical view renders the mapping of process view and module view to the underlying hardware. There is a large body of literature on software architecture which cannot be discussed here. In the following, some architectural alternatives, i.e. layered, centralized, peer-to-peer and agent architectures, are briefly described which are needed to understand the EKI architecture presented in section 1.6.3. Layered architecture. A common principle to structure software systems is to form layers. One goal is to master complexity by minimizing dependencies between layers so that changes on one part of the system do not affect other parts. Another goal is to design systems with interchangeable layers. A general rule of a layered architecture is that components on one layer can access each other without restrictions, whereas interaction across layers is restricted by (standard) interfaces (Figure 1-12). Advantages of a layered architecture are that software or its components can be reused, maintained, standardized, ported and tested separately. Disadvantages are that extra layers can harm performance because data has to be passed through multiple layers which may lead to substantial overhead. However, encapsulation of functionality in components within layers increases reus-

Architectures

Views on software architecture

Section overview

Layers

48

1. Foundation

ability and reduces maintenance efforts that often compensate for additional efforts induced by communication between layers. component 2.1

component 2.2

component 2.3

layer 2

component 1.1

component 1.2

component 1.3

layer 1

component 0.4

layer 0

component 0.1

component 0.2

component 0.3

Figure 1-12. Generalized version of a layered architecture

The ISO/OSI-layer model (section 2.1.3, 89ff) is an example for a layered architecture. Changes in lower layers, e.g., using Ethernet instead of Token Ring on the network layer do not affect higher layers. Thus, it is possible to operate TCP/IP on different lower-level network layers.

2-tier architecture

2-tier vs. 3-tier architecture. The use of layers as a principle to structure software became more and more apparent with client-server systems during the late 80s and 90s of the last century. Those systems usually had 2-tier architectures. In this configuration, a client contains application and presentation logic and a server offers a (relational) database (Figure 1-13). This architecture is well-suited for data-intensive applications to display and update data. However, more complex logic, such as calculations or validations of data, has to be implemented either on the data source layer or on the presentation layer. Choosing the data layer leads to the danger of application logic becoming dependent on data structures. If data structures have to be changed, application logic also needs to be updated. Implementing business logic on the presentation layer means that it has to be duplicated when alternative ways of presenting data are required. 2-tier architecture presentation & domain logic

data source logic

3-tier architecture presentation logic domain logic data source logic

Figure 1-13. 2-tier architecture vs. 3-tier architecture 3-tier architecture

In a 3-tier architecture, business logic is encapsulated into a separate layer (Figure 1-13). Today, systems with a 3-tier architecture are fairly

1.5. Information System Architectures

common. The need to access data with Web interfaces, e.g., using standard Web browsers, and propagation of object-oriented languages like Java may have accelerated this development. The characteristics and tasks of each layer are as follows: Data source logic is responsible for communication with other systems that manage data on behalf of the application. Examples are transaction monitors, other applications or messaging systems. The most important data sources for enterprise applications certainly are database systems responsible for storing persistent data. Domain logic performs transformations on data required to represent business transactions, i.e. the domain. Thus, it is also referred to as business logic. It comprises, e.g., calculations on the basis of inputs and stored data or choices of data sources, depending on commands received from presentation logic. Presentation logic handles interactions between users and application. Its task is to display information to the user, interpret user commands and translate them into actions performed by domain and data source layers. Graphical user interfaces are most common these days, text-based presentation and command lines are alternatives that have more and more receded. Today, they are mainly applied for specialized tasks like maintenance and administration of computer systems. Principally, users can be humans or other systems. The general principle is that all layers should be independent from each other as much as possible. For instance, if sales data needs to be aggregated for a defined set of regions, it cannot be done in the presentation layer. In the short-term, this would not be a problem, but if the functionality is needed later, e.g., to print the values, the algorithm has to be duplicated. This might lead to inconsistencies and increase efforts required for maintenance. Centralized client/server architecture. A variety of ICT services in an organization needs to be offered for and shared by all or at least a considerable number of users of the infrastructure to ensure, e.g., communication and efficient use of resources. Thus, it is beneficial to offer some services centrally by so-called servers. The units that request and use these services are called clients. In the context of ICT infrastructures, these terms are usually used for logical (software) units like applications, software components as well as single software objects. However, they are also applied to denote hardware devices, i.e. a server computer is a (dedicated) machine which solely provides services to (users on) other machines, particularly personal computers or work stations which are called client computers. Different types of clients can be distinguished depending on how work is split between client and server, i.e. which layers are performed on the client and which ones are on the server. Clients that only contain presenta-

49

Data source logic

Domain logic

Presentation logic

Example

Clients and servers

Thin clients vs. rich clients

50

1. Foundation

Rise of thin clients

tion logic are called thin clients. Clients that additionally contain (substantial parts of) domain logic are termed rich or fat clients (section 5.2, 374ff). In some settings, all three layers might be included on a client, e.g., to ensure fast responses and offline operation. Client and server then usually need to offer synchronization functions to ensure data consistency. From the view of administration and maintenance, thin clients are often most desirable as this reduces efforts for software and hardware maintenance. This is one reason for the rise of Web-based systems within organizations. However, thin clients restrain usability and interaction with applications, e.g., through less responsiveness due to limited computing power and network bandwidth as well as restricted possibilities for user interactions. Table 1-5 lists typical services centrally provided by servers. Table 1-5.

application area service

examples

identification

addressing

domain name system (DNS), dynamic host configuration (DHCP)

authentication

LDAP directory, issuing/checking of certificates

scanning

network scanner

printing

network printer

communication

mail, teleconferencing, remote access

application

application server

computing

computation of chemical structures or flow models

Web provisioning

Web server

filtering

virus scanner, SPAM filter

privacy, restricted usage

digital rights management, encryption

caching

Web proxy

data storage

file server, relational DBS, document management system

input / output

processing

security

storage

synchronization Implementation

Examples for typical server services

backup

tape library

time

network time protocol

Technical issues that constrain decisions about how centralized services are implemented, e.g., type and scope of services, number of served clients, are network bandwidth and performance of server hardware. Large

1.5. Information System Architectures

server systems might include several machines that focus on tasks such as database services, processing services and Web services. If one server fails, another one can act as backup, called failover. Usually, this is not visible for users, the configuration of servers appears as to be one system to them. Load balancing is used to distribute requests from clients to server resources. They are especially applied, if the number of requests cannot easily be forecasted. For example, already in 2004 the so-called Googleplex of the Internet search engine Google (www.google.com) was supposed to be composed of in total more than 30,000 machines in different locations. In Dublin, one of Google’s European locations, the company hosted a minimum of 100 racks each with 88 CPUs (dual 2 GHz Intel Xeon, 2 GB of RAM, 80 GB hard disk; all figures from April 2004). This hardware ensures quick responses to search requests and up to date search results due to frequently crawling and indexing Web sites. Peer-to-peer architecture. The term peer-to-peer denotes the idea of a network of equals (peers) that provide resources such as CPU time, storage area, bandwidth or data to each other without a central server. In the terminology of client/server-systems, a peer acts both as client and as server: Every peer uses services of and provides services to other peers. Together, peers shape a dynamic, continually changing network with each node always being able to participate in or leave the network and to choose autonomously which other nodes to communicate with. The peer-to-peer idea is not new, it is one of the oldest ICT and telecommunication architectures with telephone system, Usenet and Internet as major examples employing this metaphor. However, due to the tremendous growth of peer-to-peer file sharing networks, the idea has risen again on individual personal computers as opposed to backbone networks. Ideally, peer-to-peer networks can be described by the following characteristics: mutual client-server-functionality: every peer can act as a client and as a server, thus rendering all nodes functionally equal, no central authority: there is no central node which coordinates communication between peers, autonomy: peers are solely responsible for their activities, especially for determining what resources they share when and with whom, distributed data: every peer stores a fraction of the whole system’s data and there is no peer that manages the whole data, local interactions: every peer only sees its direct neighborhood, i.e. the peers it is directly connected to, dynamic changes: peers and connections between peers are not necessarily reliable which means that they may go off-line or yet cause damage to the network by spreading viruses or worthless contents.

51

Failover, loadbalancing

Example: Google

Peer-to-peer, a new idea

Characteristics of peer-topeer networks

52

1. Foundation

Differences to client/serverarchitecture

A main difference to client/server systems is that the number of peers strongly determines the overall amount of resources available. Every participating peer adds resources like storage capacity and processing power to the network. This is the reason why peer-to-peer systems are often characterized as being highly scalable. However, network bandwidth may limit the size of a peer-to-peer network since a considerable share of network traffic is necessary to discover other peers and their resources. Another important difference especially interesting for KM is that peer-to-peer networks are designed to easily form ad-hoc networks, which is a good foundation for spontaneous collaboration and knowledge sharing. Examples for application areas of existing peer-to-peer networks are: instant messaging, e.g., the ICQ network, IP telephony, e.g., the Skype network file sharing, with prominent examples, e.g., Gnutella, Kazaa, Napster or Bittorrent, i.e. peer-to-peer software that supports the sharing of files in networks of users, distributed and grid computing which aims at coordinated usage of distributed computing power, e.g., search for extraterrestrial life (SETI@home used >3 million computers to achieve 23.37 sustained teraflops in September 2001; still >2.1 million hosts running in 2008), or protein folding (Folding@home had achieved peaks of 1502 teraflops on over 270,000 machines as of March 2008), collaboration, with Groove (section 4.3.5, 301ff) being the most cited peer-to-peer collaboration platform. However, there are still serious technical challenges that have to be overcome in peer-to-peer computing. These challenges concern: connectivity, e.g., locating peers that do not have public IP addresses and mechanisms for communicating through firewalls, security and privacy, e.g., risk of spreading viruses, unauthorized access to confidential data or installation of unwanted applications, fault-tolerance and availability, e.g., finding the required resources available when they are needed, scalability, especially concerning the naming scheme and searches in the flat structure of the distributed search domain, self-managed systems that are administered by individual users with limited experience and tools who provide services to others and interoperability, i.e. current peer-to-peer installations cannot connect to each other due to, e.g., a variety of computing models, a variety of network settings and a wide range of application types. In the context of organizations, peer-to-peer systems should rather be seen as complementing existing client-server systems than as competing approach. Many tasks are immanently centralized. Examples are processing and analyzing large amounts of business transactions with manage-

Application areas

Challenges

Ideal types of peer-to-peer architectures

1.5. Information System Architectures

53

ment support systems and maintenance tasks like software license management, software updates or bandwidth management. Even peer-to-peer systems might deploy centralized devices, e.g., for authentication, indexing, message relaying and discovery of other peers. Contrary to the clientserver architecture, peers communicate and offer services directly to other peers. Three ideal architectural types of peer-to-peer systems can be distinguished with respect to existence of any central assistance or the role that central assistance plays in the network (Figure 1-14):

peer super peer server pure peer-to-peer architecture

assisted peer-topeer architecture

super peer-topeer architecture

Figure 1-14. Ideal architectural types of peer-to-peer systems

The pure peer-to-peer architecture does not have any central authentication or coordination mechanism. Every peer provides complete client and server functionality. The assisted peer-to-peer architecture requires a central server, e.g., to authenticate users, to discover other peers, to act as global search index, to relay messages or to provide interfaces to other centralized systems. The super peer-to-peer architecture is in between assisted and pure architectures. A peer is connected to one single super peer, thus forming clusters of peers in the network. Super peers are also connected to each other, thus forming a hierarchically superordinated peer-to-peer network. Requests from peers are always handled by the connected super peer and eventually forwarded to other super peers. As in the assisted architecture, a direct connection between peers is established, once a peer with the desired resource is found. Agent architecture. In general, software agents should not be seen as alternatives to centralized or distributed systems, but as complements. In the literature there is no single accepted definition due to heterogeneous research domains of the community of agent researchers. There is a widely accepted set of attributes, though, that a piece of software must have in order to be considered an agent. The most important attributes are autonomy, rationality and pro-activeness.

54

1. Foundation

Definition of software agent

A software agent is an autonomous software entity that can act reactively and proactively, can communicate with other agents and uses a rational calculus to develop and carry out plans in order to reach its goals (Wooldridge/Jennings 1995, Kirn 2002).

Further characteristics of agents

Many other attributes like intelligence, sociality, intention, deliberation, cognition, mobility and persistence are often used in conjunction with agents, but most of them are either not essential (mobility and persistence), difficult to define (intelligence, sociality) or are just other words for aspects already covered by autonomy, rationality and pro-activeness. The definition above is sufficiently general to cover a large number of concrete software systems, starting from Unix processes (daemons) that run in an own thread, are monitoring their environment and communicate with other processes by passing messages or invoking functions. Processes of this kind are regarded as representing the weak notion of agency, i.e. a wide interpretation of the term. A narrow interpretation, called the strong notion of agency, is mainly represented by AI research demanding that agents have to have sophisticated mental models, use machine learning algorithms and symbolic reasoning. Agents must not be mixed up with objects (in the sense of object-oriented programming). Objects encapsulate data and methods that operate on this data. Agents also encapsulate their behavior. That leads to the fact that often an agent exposes only a very limited set of generic methods to its environment that hide its real purposes. Examples for methods are perceive(), reason() and act(). In general, agents exist in an environment which is monitored by them. They have an internal representation of the environment (world model) and interact with it. The main distinction between agent architectures is how they are doing this. There are a large number of suggestions for agent architectures for general or for specific purposes. They can be classified as reactive, cognitive and deliberative architectures. Reactive agents only wait for certain stimuli and react in one of certain pre-defined ways based on their internal state. Therefore they do not have to make plans for the future. Cognitive agents are influenced by psychology so that their architecture is often an implementation of a model of human mind. Examples are Soar and ACT-R. Deliberative agents finally contain a symbolic world model and decide based on logical reasoning, pattern matching and symbolic reasoning. There are also hybrid architectures that integrate reactive and deliberative behavior as humans also have reflexes for quick reactions in critical situations and reasoning for decisions with long reaching consequences. One example for a hybrid architecture is InterRAP (Müller 1995). It is a layered architecture with three layers (Figure 1-15): the behavior-based layer for reactions,

Agents and environment

Reactive architecture Cognitive architecture Deliberative architecture Hybrid architecture Example: InterRAP

1.5. Information System Architectures

55

the local planning layer for plans the agent can carry out on its own and the cooperative planning layer for plans that involve activities of other agents. agent knowledge base

agent control unit

social model

situation recognizer, goal activator

planning and scheduling

cooperative planning layer

mental model

situation recognizer, goal activator

planning and scheduling

local planning layer

world model

situation recognizer, goal activator

planning and scheduling

behaviourbased layer

action

world interface, body

perception

communication

environment

Figure 1-15. InterRAP agent architecture

Accordingly, the knowledge base of an InterRAP agent is also divided into three parts: (1) world model for reacting, (2) mental model for local planning and (3) social model for cooperative planning. Layers are activated bottom-up. If the lower layer cannot cope with a given situation, then a higher layer is activated. For planning, InterRAP builds on the BDI model (Belief, Desire, Intention, Rao/Georgeff 1991). Beliefs are statements about the environment, the agents think are true. The sum of all beliefs represents the world model of an agent. Desires are states of the environment and the agent itself that the agent wants to achieve. Desires often are contradicting so that goals have to be set that are consistent and reachable. Intentions are concrete actions that an agent wants to take in the near future to reach a goal. There are a lot of application areas for software agents like distributed problem solving and controlling complex machines. For knowledge infrastructures, two types of applications are of special interest: information agents and interface agents. Information agents monitor information sources and collect information they consider useful for their user, often based on keywords that represent the user's interests. They help a user to stay up to date without the need to

BDI model

Application in knowledge infrastructures Information agents

56

Interface agents

1. Foundation

manually select relevant information. Typically, the agent uses extensive user profiles that are permanently updated to get more relevant results than usual search engines. It typically also acts as a meta-search engine and Web crawler by retrieving information from several search engines and additionally crawling pages that are not indexed by search engines. Examples of such agents are Copernic Agent and WebMate. Interface agents monitor user activities and suggest improvements, provide help for functions or automate short tasks. They often have graphical representations that help users to develop a sort of “social relationship” with the machine. An example for a simple interface agent is the MS Office assistant. It acts as a shortcut to the applications’ help functions and provides hints thought helpful in the current situation. In the context of knowledge infrastructures, an interface agent could monitor the user’s activities, deduce the mode and context a user is working in and suggest a number of meaningful actions the user could take (section 6.1, 408ff).

1.5.2 Enterprise Architectures

Business information systems

Classification of BIS

Knowledge work was described as creative work that requires creation, acquisition, application and distribution of knowledge, but also data and information as inputs and outputs (section 1.2, 21ff). Individuals concerned with knowledge and information work deal with structured and semi- or unstructured data. EKI is primarily targeted at semi- or unstructured, contextualized information, but also needs to integrate organizational systems that manage structured data. These systems have a long tradition in representing business transactions and the flow of materials, goods and services through an organization in ISs. They have been termed operational or transaction processing systems and store representations of transactions in database systems where the resulting data can be filtered, aggregated and integrated into data warehouses. Data is then analyzed and presented with the help of MISs, data mining and decision support systems as well as executive ISs. Altogether, these systems focus on transactional and thus often structured data and are called business information systems (BIS). From the perspective of an EKI, BIS primarily play the role of data sources. They can be integrated at different levels (data, domain or presentation layer, section 1.5.1) depending on the requirements of the actual application scenario. This section gives an overview of BIS for managing structured data with the help of two schemes that are suited to form classes of IS: the IS pyramid and the value chain. IS pyramid. This scheme classifies systems along two dimensions: (1) the time frame of decisions they are designed to support and thus the time horizon of the corresponding data and (2) the business or functional area in which the system is applied (Figure 1-16).

1.5. Information System Architectures

57

e

procurement IS

procurement controlling

tm en t IS

inves

inves tm con tr en t olling

pro co d uct i ntr oll on ing

management information systems (MIS)

urc eso n r ling ma ol hu ontr c

technology

ing

production

accounts current creditors

sales

warehouse asset accounting accounting

con troll

vertical integration

n S ma s I hu ur ce o r es and salesution IS distrib

pro du ct I S ion

executive information systems (EIS)

accounts current debitors

procurement

sales

humanresource accounting

office information and accounting systems

humanresource allocation

transaction processing systems

horizontal integration

Figure 1-16. Information systems pyramid (based on Scheer 1994)

The considered time frame can be: operative, i.e. with a short time horizon (hours, days or a few weeks) which means they are less important for the long-term development of the organization, e.g., reordering of consumable materials, tactical, i.e. with a medium time horizon (several weeks up to two years) and are placed to steer the operative business, e.g., choice of supplier for the next year, strategic, i.e. with a long time horizon (two to five years) constituting the development of the organization and its goals, e.g., definition of product lines based on the selection of product-market combinations. The functional areas are classified according to the main business functions, e.g., purchasing, materials management, production, marketing, sales or human resource management (HRM). The business functions are supported by specific software functions, modules or entire systems, e.g., e-procurement system, materials management module of an ERP system, production planning system (PPS), computer-integrated manufacturing (CIM) system, sales management system, customer relationship management (CRM) system, HRM module of an ERP system. The IS pyramid shows two directions for integration between systems. Horizontal integration connects systems of different organizational functions on the same level. Vertical integration connects systems within the same organizational function on different levels. In this case, integration

Time horizon

Functional area

Horizontal and vertical integration

58

Transaction processing system

Accounting system

Office information system

Management information system

Executive information systems

Primary and support activities

1. Foundation

means connecting data and coordinating tasks, processes, methods and programs (chapter 3, 151ff). The following IS classes can be distinguished according to the different levels of the pyramid: Transaction processing systems (TPS, also called operative IS) support day-to-day operations with the aim to increase efficiency of operative processes. They manage data about quantities of, e.g., parts, products or service units on a fine-grained level. Data is handled by (relational) database management systems for flexible, application-independent and efficient storage and retrieval of data. Accounting systems are oriented towards financial value. These are systems to support business transactions that handle data about payments, billing and accounts. Data about single operative tasks is aggregated to business transactions. Nevertheless, accounting systems still handle data on a low level of granularity. Office information systems (OIS) support administrative tasks by helping employees to collect, transform, store and exchange data. Examples are text processing software, communication and collaboration applications (section 4.3, 285ff) and project management software. These enduser tools are sources for semi-structured and unstructured data such as text documents, e.g., proposal, meeting protocol, spreadsheets, e.g., budget plan, work time record, and presentations, e.g., product information. Management information systems (MIS) assist lower, middle and upper management in planning and control tasks, e.g., by analyzing business data, comparing targets and actual business results, guiding and structuring decisions as well as giving hints and suggestions for corrective measures. Examples are decision support systems (DSS), query and reporting systems, online analytical processing (OLAP) for analysis and exploration of business data. Many of these tools build on data warehouses that manage a database of theme-oriented, integrated, time-variant and non-volatile data in support of management’s decisions. Executive information systems (EIS) finally are ISs that target senior and top management of an organization. EIS highly aggregate data to only a few measures that often employ advanced visualization techniques with well-known metaphors to direct attention to crucial developments or exceptional data constellations, e.g., portfolios, traffic lights, cockpits, maps in geographical ISs or warning thresholds. IS along the value chain. The value chain structures organizations’ activities to create value and is an instrument to analyze sources of competitive advantage (Porter 1985). Value is defined as output of organizations that customers are willing to pay money for. Value is created along the value chain through the contribution of single value-creating activities. Value-creating activities can be classified into primary and support activities. Primary activities are directly related to value creation, e.g., inbound logistics, operations, outbound logistics, marketing and sales, as

1.5. Information System Architectures

well as customer service. Support activities facilitate primary activities and other support activities by supplying resources such as human resources, infrastructure and technologies. Value creating activities are the foundation for an organization to create a margin that results in profit. The value chain arranges value-creating activities on a general, abstract level. On a more detailed level, activities can be described by business processes that sequence actions, functions or tasks as parts of activities. In analogy to the classification of activities in primary and supporting activities, core processes and supporting processes are distinguished. Classes of IS can be classified according to activities and business processes along the value chain which determines their data types and domains. Figure 1-17 depicts Porter’s value chain and related examples for system classes. In the following, the discussion concentrates on those classes of systems that have not been presented yet, starting with systems that facilitate support activities of an organization’s value chain. support activities

technology development

research & development applications

human resource management

inbound logistics

operations

human resource management systems

customer relationship management systems

marketing & sales

customer service

shipping systems

outbound logistics

ma rgin

supplier relationship management systems

rgin ma

e-procurement systems

procurement

computer integrated manufacturing

Business processes

facility management systems

firm infrastructure

supply chain management systems

59

primary activities

Figure 1-17. Examples for information systems along the value chain

Facility management systems ease planning and control of the organization’s basic infrastructure, i.e. buildings, premises and related basic services such as electricity, illumination, air conditioning or network access. This supports tasks such as managing usage requests, e.g., booking of meeting rooms, machines or vehicles, accounting usage expenses, tracking maintenance cycles and controlling basic services. Generally, e-procurement supports cost-, time- and quality-efficient acquisition of goods and services ordered by single employees by means of electronic systems. The term is often used in a narrower sense to denote the ordering of maintenance, repair and operations (MRO) goods directly

Facility management systems

E-procurement systems

60

Research & development applications

Human resource management systems Supply chain management systems

Computer integrated manufacturing

Customer relationship management

Shipping systems Supplier relationship management Enterprise resource planning

1. Foundation

by the consumer, i.e. an employee at the workplace. Important motivation for the introduction of e-procurement systems is that centralized procurement of cheap goods can cause disproportionally high costs. Systems supporting research and development (R&D) in organizations are highly specialized according to specific domains of research or industry sector. This comprises systems that support planning, development and simulation of, e.g., new services, products and technologies such as computer aided design (CAD) applications, expert systems, visualization tools and simulation systems. HRM systems support and automate selected tasks related to managing personnel, e.g., recruiting, calculating and paying salaries, planning career paths and skill development as well as managing an employee’s or a group’s workload. Data security and protection are important issues as the stored data is very sensitive. SCM systems aim at increasing efficiency of processes associated with acquiring goods and services from a set of suppliers with long-standing relationships, managing inventory, and supplying and processing of materials. Examples for systems that facilitate SCM are EDI (electronic data interchange) systems for standardized data exchange between customer and supplier and collaborative planning systems for supporting a joint solution of planning problems by members of different organizations. SCM operates on structured data about products, services and processes of suppliers and data of inventory management. Main aim of CIM is integrated support of production operations and of disposition and control tasks to facilitate efficient operations. Goals are, e.g., optimal utilization of resources, optimized stock of inventory, high product quality, compliance with delivery dates and short cycle times. PPS support materials and capacity management as well as production control. CRM systems comprise techniques and software to manage data describing customers, business transactions and communication acts with customers in an organized and efficient way. CRM systems integrate systems and tools that contain detailed customer information, e.g., offering systems, help desk systems, call center, and marketing ISs, and support all tasks for marketing, sales and customer service, e.g., to match customer needs with product offerings and offer individualized customer services. Shipping systems support all aspects related to product or service orders, delivery and billing. This includes management of freight carriers, auditing of orders and bills, freight tracking and accounting tasks. Supplier relationship management (SRM) is similar to CRM with the difference that it is directed to the supplier side. It deals with financial information, technologies, research and development as well as future developments of suppliers. Many of these system classes are incorporated by ERP systems. The term is commonly used for standard software that integrates and automates tasks associated with all operations or production aspects of an organiza-

1.5. Information System Architectures

61

tion. Traditionally, ERP systems were seen as an extension of manufacturing resource planning (MRP II) systems. ERP systems support manufacturing, logistics, distribution, inventory, shipping, invoicing, accounting, human resources, and marketing. They are sources for fine-granular operative data representing all areas of operative business activity. Well-known product vendors are Microsoft, Oracle and SAP. Integrated enterprise architecture. Integration denotes (re)construction of a systematic whole. One reason for the need for integration is specialization. Though it is necessary to efficiently accomplish tasks and utilize resources, it often results in unnecessary and impeding boundaries between functional areas, processes or organizational units. Enterprise application integration (EAI) comprises methods and technologies for integrated support of business processes within and between organizations by enterprise systems. It is a foundation for inter-organizational activities like e-procurement, SCM and collaborative business networks. Many monolithic, single-purpose systems (legacy or “stove-pipe” systems) are not replaced because they are often the backbone of the organization’s operative business transactions and it is considered not feasible or too risky to replace them. Thus, they have to be deployed and maintained together with newer systems. Additional reasons for integration can be seen in: Mergers and acquisitions: They usually cause the need to integrate on different levels, e.g., in terms of strategy, personnel and IT infrastructure. Depending on characteristics of IT systems and other context factors, alternatives are to completely supplant ISs of one partner, to work independently with different IT infrastructures or to connect both infrastructures on data, application or user interface level with EAI. Cooperations: Different types of electronic communication and collaboration mechanisms are employed to realize, support or improve interorganizational business and knowledge processes. Typically, an organization's systems are connected to external processes or data sources, because business activities are realized jointly with business partners. Business process improvements: Integration of ISs often provides potentials for improving business processes. Integrated enterprise architectures built, e.g., with the help of ERP systems, and automation can help to significantly improve internal business processes. Success of ERP systems depends on the joint support of internal business processes, either by providing comprehensive functions or by connecting to other internal systems. Automation of processes usually makes it necessary to integrate tasks or process steps on a technical level. The potential advantages of integration are: Higher flexibility. An organization can adapt more easily to changing business environments, unforeseen changes and different customer

Enterprise application integration

Needs for integration

Advantages of integration

62

Data layer

Domain layer

Presentation layer

1. Foundation

needs by faster reconfiguration of its business processes and the software that supports these processes. Cost reduction. Processes are automated and unnecessary activities are avoided, e.g., generation of information already stored in other systems. Better quality. Quality of services and products increases in terms of lower response times, fewer errors through manual data input or better decisions through updated information. Time savings. Response or production times are shorter because internal processes can be automated and supplier or customer processes are integrated into the company's processes more tightly. It simplifies development of new software or extensions of existing software and reduces the complexity of the development process. Integration can be realized on all three layers of software architecture, i.e. data, domain and presentation layer. This depends on the actual challenges, characteristics of the infrastructure and capabilities of developers and IT personnel. The goal of integration on the data layer is to build one common database providing “master data” for applications which formerly stored their data separately. Separate data storage often leads to data inconsistencies due to the danger that data entered or changed in one system is not updated simultaneously in all relevant applications. Another result of integration on this level can be to better understand an organization's data which can lead, e.g., to the identification of unused cross-selling-potentials. Integration on the domain layer targets integrated support of business processes that involve multiple applications. Coordination of applications and exchange of data between them should be automated as much as possible. Another goal of application-level integration is to reuse software components (section 2.3.3, 137ff). One approach to implement domainlevel integration is to insert an additional software layer to enable communication between applications, called middleware. Integration on the presentation layer is the most basic type of integration and often the “last resort” if integration on lower layers is not possible, i.e. if no interfaces exist to access an application’s data or its functions. It aims at presenting different applications or systems in a more or less consistent and standardized way to the user. The most prevalent approaches are screen scraping and Web integration using Frames or a Portal server. Screen scraping is a common way to integrate text-based interfaces of legacy systems. These terminals either are displayed directly with aided by emulators or indirectly by interpreting data streams from mainframe to terminal to present it in a more user-friendly way. Web integration organizes the output of multiple systems in HTML pages displayed with Web browsers. This could be done rather simply, e.g., by displaying each system within a separate browser frame, or more advanced by using dedicated Portal servers (section 5.1.1, 366ff).

1.5. Information System Architectures

Example: SAP Enterprise Services Architecture. By the end of 2003, the German software company SAP decided to undertake steps for reducing software maintenance and development costs of SAP ERP, a widelyused standard software product suite for enterprise resource planning based on the client/server model. SAP NetWeaver is an integrated application platform and foundation for development of an enterprise system by providing a basic infrastructure for interaction between software components. Its architecture reflects the importance of integration issues (Figure 1-18). According to the architecture, integration happens at four levels: SAP NetWeaver interfaces

people integration multi-channel access

2

portal

3

business intelligence

2

collaboration

information integration 2

knowledge management

4

master data management

5

process integration

integration broker

6

business process management

Web Sphere solution life cycle management

composite application framework

1

...

...

application platform J2EE

ABAP

data base and operating system abstraction

...

Figure 1-18. Architecture of SAP NetWeaver (SAP)

people4, i.e. integration of internal and external persons with the help of multi-channel interfaces and by e.g., portals (section 5.1.1, 366ff), as well as by tools for collaboration, information, i.e. integration of structured data from operative systems handled by master data management which can be analyzed by means of business intelligence functions as well as integration of unstructured data handled by functions called “knowledge management” by SAP,

4

Please note that people are not really integrated into the architecture. It is an allegory used for integration of access to services and data for people.

63 SAP ERP

SAP NetWeaver

64

Composite application framework, xApps Solution lifecycle management

Critique

1. Foundation

processes, i.e. integrated support of internal processes and connection with external processes and workflows, e.g., of suppliers (section 3.4, 219ff) and applications, i.e. an integrated application platform supporting multiple programming languages, operating systems and database management systems with the capability to expose functions as Web services. The composite application framework comprises tools, methods, rules as well as patterns and supports developers (of SAP or of solution partners) to create so-called packaged composite applications that use bundles of Enterprise Services for individual business scenarios and business processes. SAP offers pre-configured applications for business processes. The solution life-cycle management provides functions to manage software that is part of the NetWeaver architecture, i.e. installation, upgrade, monitoring, customizing and testing of software. NetWeaver provides interfaces to systems implemented in other programming languages and from other product vendors. This architecture reflects a view on integration that is strongly focused on products currently provided (or planned to be provided) by SAP. It should be noted that it does not reflect the perspective on integration of EKI, which will be discussed thoroughly in chapter 3.

1.6 Knowledge Infrastructures The final section of this chapter draws together the theoretical foundations discussed so far and develops the concept of EKI. Section 1.6.1 first provides a definition of EKI, before section 1.6.2 deals with the central element of an EKI, the concept of a knowledge or KM service. Finally, section 1.6.3 presents the EKI architecture which also provides an outline for the subsequent chapters of this book.

1.6.1 Definition EKIs are implemented as part of a KM initiative that comprises a number of KM instruments. After discussing knowledge, knowledge work, management and instrument, this section relates these concepts to EKI with the help of five perspectives on KM as well as a number of distinctive characteristics of EKI that ultimately lead to a definition of the term EKI. Main modeling concepts

KM perspectives. The design of KM initiatives requires joint consideration of the most important KM modeling concepts. These are (1) KM instruments, i.e. strategic arrangements of measures for systematic interventions into handling of knowledge, (2) organizational design, i.e.

1.6. Knowledge Infrastructures

65

knowledge tasks, processes, roles and responsibilities, (3) persons, i.e. skills, communication and cooperation in networks and communities, (4) knowledge products and structures, i.e. topics or domains, types of knowledge, meta-data and ontologies as well as (5) ICT tools and systems, i.e. functions, architecture, structure and interaction of EKI (Figure 1-19). motives, outcome

strategy, measures occasion

deployment

sub

event, condition, action

a in dom

ject

(1) instrument

goal, input, output

netw c om ork/ mun ity

(3) process res e ou rol r ce meta-data y t i l performance ap crea communication scope t i bi i p o s lica n/ support n o p tio n re s expert (2) person (4) product pr e of nc ile re r organizational theme cu pe navigation oc ontology rs o structure na structure t liz ati ten n on co (5) enterprise knowledge infrastructure service/ interaction

architecture/ integration

Figure 1-19. Perspectives on knowledge management

The five main perspectives instrument, person, process, product and enterprise knowledge infrastructure, are connected by a number of concepts. Persons are involved in processes by responsibilities for activities and roles that are assigned to actions. Processes, especially activities and actions, are supported by EKIs in order to improve organizational performance. Also, processes can be used to aid navigation in EKIs. Products are mapped to occurrences, e.g., documents or other resources that are stored in EKIs. Taxonomies and ontologies can be used as primary structures of EKI contents. Persons hold skills that are structured as products. Persons and networks have interests in products. Experts, e.g., subject matter spe-

Connecting concepts

66

1. Foundation

cialists, take care of certain products in organizations. Processes and products are connected by knowledge resources, both in the form of skills and in the form of documents, that are required in activities and actions and by the two fundamental tasks creation and application of knowledge. Profiles and personalization techniques are used to model persons to support their use of EKIs. Finally, instruments link the context of persons, products and EKI to steps in processes which provide occasions for knowledge tasks. Knowledge management tools and systems

Learning management systems

Enterprise knowledge infrastructure Initiative

Context

Characteristics. There are numerous approaches to define ICTs for KM. This is reflected by a large number of terms, e.g., knowledge infrastructure, knowledge management system, software or suite, knowledge warehouse or organizational memory (information) system. In addition to these terms denoting a comprehensive platform, many authors provide more or less extensive lists of individual KM tools that can be used to support entire KM initiatives or certain processes, life-cycle phases or tasks. Another group of software systems supports e-learning, e.g., e-learning suite, learning object repository, learning content management system, learning management platform, portal, suite or system. These platforms not only support presentation, administration and organization of teaching material, but also interaction between and among teachers and students. KMS with roots in AI, document management or collaboration and learning management systems with roots in computer-based training already share a substantial portion of functionality. EKIs are organization-wide platforms that include aspects of KMS and LMS. They offer an IT environment for collaboration, learning and handling information and knowledge to support knowledge work. Figure 1-20 summarizes six EKI characteristics described in the following. Goals are defined by the KM initiative in which the EKI is deployed. Primary goal of EKI is to increase organizational effectiveness by a systematic management of knowledge. Thus, EKI are the ICT environment for effective knowledge work, the technological part of a KM initiative that also comprises person-oriented and organizational measures targeted at improving productivity of knowledge work. The type of initiative determines the type of infrastructure for its support. KM initiatives can be classified e.g., according to strategy in human-oriented, personalization initiatives and technology-oriented codification initiatives. EKIs are applied to managing knowledge which is personalized information related to facts, procedures, concepts, interpretations, ideas, observations and judgements. Here, knowledge means information that is meaningfully organized, accumulated and embedded in a context of creation and application. EKIs primarily leverage codified knowledge, but also aid communication or inference. Thus, on the one hand EKIs might not appear radically different from existing IS, but help to assimilate contextualized information. On the other hand, the role of ICTs is to provide access to sources of knowledge and, with the help of shared context, to increase the

1.6. Knowledge Infrastructures

67

breadth of knowledge sharing between persons rather than storing knowledge. Contextualization provides semantic links between explicit, codified knowledge and persons that hold or seek knowledge in certain domains. Context enhances the simple “container” metaphor of organizational knowledge by a network of artefacts, people, memory and processing.

initiative process

participant context

instrument

service

platform Figure 1-20. Characteristics of enterprise knowledge infrastructures

EKIs are installed to support and enhance knowledge-intensive processes, tasks or projects of, e.g., knowledge creation, organization, integration, storage, retrieval, transfer, refinement and packaging, (re-)use, revision and feedback, also called the knowledge life-cycle, ultimately to support knowledge work. In this view, EKIs provide a seamless pipeline for the flow of (explicit) knowledge through a refinement or maturing process (section 1.4.4), a forum for interpretations, half-formed judgements, ideas and other perishable insights that aims at sparking collaborative learning. Users play the roles of active, involved participants in knowledge networks and communities of knowledge workers that “own the knowledge” and decide what and how to share fostered by EKIs. Meta-knowledge, e.g., in the form of a set of expert profiles or contents of a skill management system, is sometimes as important as the original knowledge. Context is again needed here to provide semantic links between codified knowledge and people or collectives as holders of knowledge, between handling of explicit and implicit knowledge and between documented knowledge and meta-knowledge, feedback, valuations and comments about applying knowledge elements by other participants respectively. EKIs are applied in many application areas, e.g., in product development, process improvement, project management, post-merger integration or HRM. More specifically, EKIs support KM instruments, e.g., (1) capturing, creating and sharing best practices, (2) implementing experience management systems, (3) creating corporate knowledge directories, (4) expertise locators, yellow and blue pages as well as skill management systems, also called people-finder systems, (5) collaborative filtering and handling of interests used to connect people, (6) creating and fostering

Process

Participant

Instrument

68

Service

Platform

1. Foundation

communities or knowledge networks, (7) facilitating intelligent problem solving. EKIs offer a targeted combination and integration of knowledge services that together foster selected KM instrument(s). Integrated services are built on EKIs as ICT platforms. The supported processes, participants and instruments determine the types of services needed. Examples include rather basic services, e.g., for collaboration, workflow management, document and content management, visualization, search and retrieval as well as more advanced services, e.g., profiling, personalization, text analysis, clustering and categorization to increase the relevance of retrieved and pushed information, advanced graphical techniques for navigation, awareness services, shared workspaces, (distributed) learning services as well as integration of and reasoning about various (document) sources on the basis of a shared ontology. Whereas the focus on participants represents a user-centric approach, an IT-centric approach provides a base system to capture and distribute knowledge. This platform is then used throughout an organization. In this view, EKIs are not application systems targeted at a single KM initiative, but can either be used as-is to support knowledge processes or as integrating base system on which specific KM application systems are built. Comprehensive means that the platform offers extensive functionality for user administration, messaging, conferencing and sharing of (documented) knowledge, i.e. publication, search, retrieval and presentation. Definition. A definition of the term EKI has to focus on (1) a platform rather than individual KM tools. This platform targets (2) collaboration, sharing of information and knowledge as well as learning, contains (3) knowledge services for investigation, individuation, interaction, in-formation and instruction (section 1.6.3) based on (4) integration services that contain an enterprise-wide base ontology or ways to map ontologies. Contents and services are (5) personalized for users playing active roles in knowledge networks. Combining and integrating services realizes ICTsupported (6) KM instruments around which content and context of these systems revolve. The entire life-cycle of (7) processes of acquiring and deploying knowledge has to be established. Also, EKIs have to be aligned with the specifics of its application environment, goals and type of a KM initiative and finally aims at (8) productivity of information and knowledge work.

1.6. Knowledge Infrastructures

69

An enterprise knowledge infrastructure is (1) a comprehensive ICT platform (2) for collaboration, knowledge sharing and learning (3) with knowledge services built on top that are (4) contextualized, integrated on the basis of a shared ontology and (5) personalized for participants networked in communities (6) that fosters the implementation of KM instruments (7) in support of knowledge processes (8) targeted at increasing productivity of knowledge work.

Definition of enterprise knowledge infrastructure

Actual implementations of ICT systems certainly fulfill the characteristics of ideal EKIs only to a certain degree. Thus, a continuum between traditional ISs and advanced EKIs might be imagined. Still, many ICTs are available on the market that can be integrated to fulfill the vision of EKIs. EKI is a concept that can be seen as complementary to ERP. Both concepts represent comprehensive ICT infrastructures. ERPs focus on the informational representation of business transactions in well-structured administrative, dispositive and managerial business processes. EKIs provide a number of services to support any kind of knowledge work besides traditional business transactions and well-organized and well-supported business processes, e.g., in innovation projects, knowledge processes, knowledge-intensive business processes as well as personal handling of experiences and semi-structured data. In contrast to ERP systems, and similar to recent approaches there, EKIs are highly modular and consist of a number of building blocks that are partly common multi-purpose tools like Web or email servers and partly highly specialized knowledge-oriented tools like ontology inference or skill management systems. As in the case of ERP systems, not all parts of EKIs are equally important for a single organization, so that EKIs might have to be customized to fit perfectly.

Vision and available technologies EKI and enterprise resource planning

1.6.2 Knowledge Management Services Basic elements of EKIs are services. Since software engineering was founded as a discipline in the 70s, computer scientists have searched for ways to describe basic or advanced building blocks out of which software systems can be composed. The main advantages are reduced cost and time as well as improved quality through modularization, reusability, stability and interoperability of the resulting software systems. There are a number of different terms for building blocks, e.g., functions, procedures, modules, classes or components. Recently, service has been the central concept for a redefinition of the technical and conceptual foundation for these main building blocks from a more business-oriented perspective. Service concept. A service consists of contract, interface and implementation. It has distinctive functional meaning typically reflecting a high-level business concept covering data and business logic (Krafzig et al. 2005).

Building blocks for software

70

1. Foundation

Service descriptions provide information about: service capability: conceptual purpose and expected result, service interface: the service’s signature, i.e. input, output, error parameters and message types, service behavior: a detailed workflow invoking other services, quality of service: functional and non-functional quality attributes, e.g., service metering, costs, performance metrics and security attributes.

Service description

Definition of service

A service is an abstract resource that represents a capability of performing tasks that form a coherent functionality from the point of view of providers entities and requesters entities.

Web service

The service concept has gained popularity with the advent of a set of standards for open interaction between software applications using Web services5. A Web service is a software system, identified by a Unified Resource Identifier (URI), whose public interfaces and bindings are defined and described using Extensible Markup Language (XML). Its definition can be discovered by other software systems using a standardized directory service, Universal Description, Discovery and Integration (UDDI). These systems may then interact with the Web service in a manner prescribed by its definition, using XML-based messages conveyed by Internet-based protocols, e.g., Simple Object Access Protocol (SOAP)6. Web services are one way of implementing business and technical services in a service-oriented architecture (SOA). A SOA comprises application frontend, services, service repository and service bus which make functions available so that they can be accessed without any information about their implementation. SOAs promise more flexible and adaptable ISs. In the context of SOA, services have to be interoperable in terms of platform independence, own a network addressable interface and be dynamically discovered and accessed. The service concept has had a profound impact on EAI, on business-to-business applications and generally on the way ICT infrastructures are designed and managed.

Service-oriented architecture

5

In distributed systems, SOAs can be seen as successors of component architectures. The underlying conceptual change could also trigger a paradigm shift from a primarily production-oriented view, not only of software production, to a view that takes into account the specifics of the service sector. The service sector has experienced growth during the last decades as opposed to the production sector which has declined. There is currently an initiative led by IBM and Oracle, but also involving institutions such as the European Commission, that aim at defining a research agenda for so-called services sciences. This agenda should bring the vision of a service-led economy to the focus of a number of scientific disciplines. Thus, the service concept transcends the scientific disciplines of computer science and information systems and also involves disciplines such as management, economics or service engineering. 6 URL: http://www.w3.org/TR/wsa-reqs/, http://www.w3.org/TR/ws-gloss/, last access 2008-11-01, Alonso 2004, 124.

1.6. Knowledge Infrastructures

“SOA-enabled” organizations are called agile, on-demand or serviceoriented enterprises. These metaphors attempt to carry over SOA semantics to organizational design. This has connotations for changes in ICT’s general role in business (transforming business models), value creation (value networks), business processes (dynamically designed, net-like with emphasis on parallel processing) as well as organizational structure (service consumer-provider relationship complementing or even replacing traditional hierarchies). The remainder of this section will concentrate on the specifics of the service concept applied to EKIs (Maier/Remus 2007). KM services or knowledge services are a subset of services, both basic and composed, whose functionality supports high-level KM instruments as part of on-demand KM initiatives, e.g., find expert, submit experience, publish skill profile, revisit learning resource or join community-of-interest. These services might cater to the special needs of one or a small number of organizational units, e.g., a process, work group, department, subsidiary, factory or outlet in order to provide solutions to defined business problems. KM services describe aspects of KM instruments supported by heterogeneous application systems that can be combined into EKIs.

71 Agile businesses

Knowledge management service

Service infrastructure. Basic services can be composed into KM services. Figure 1-21 shows the main layers of a KM service infrastructure. Conceptual layer business process

KM service layer complex KM services

basic KM services

knowledge process

KM KM service KMservice service

KM service

KM service

KM service

KM KM service service

ICT layer e.g., Web services

WS3 WS2 WS1

application systems

System 1

WS9 WS4

WS9 WS4

System 2

System 2

WS10 WS8 WS5 WS7 WS6 WS5

System 3

Figure 1-21. KM service infrastructure

Based on process descriptions, the conceptual layer defines which services are required in which core business processes, which services are

Conceptual layer

72

ICT layer

KM service layer

Service composition

Service parameters

1. Foundation

offered by what service processes, who is responsible for them and what resources are allocated to fulfil them. Concepts of process-oriented KM can help to analyze, understand and design business and knowledge processes with regard to a knowledge-oriented and at the same time a strategic perspective on KM services in business processes. Services are described, discovered and invoked with the help of negotiated or standardized sets of technologies, e.g., in the case of Web services WSDL, UDDI and SOAP. These technologies support integration on different levels, i.e. human-to-machine, machine-to-machine and inter-organizational integration. The ICT layer comprises infrastructure, integration, knowledge and access services (section 1.6.3) dispersed over a variety of heterogeneous application systems. The KM service layer bridges the gap between conceptual and ICT layers. KM services have to be composed using services offered by the ICT layer. Additionally, discovery, call and provision of KM services from different activities of business processes have to be supported. It has to be determined which services are required for processes. At one extreme, a process can be viewed as one single, but complex service; at the other extreme, service granularity can be so fine that each task in a process can be constructed from multiple services. Similar to concepts in SOA, the choice is made by balancing quality of service characteristics (QoS), ubiquitous service reuse and reducing complexity for service composition. Regardless of the implementation, it is important to understand the steps required to decompose a process into a series of complex and basic services and operational characteristics. Composing KM services means specifying how these services have to be discovered and selected (discovery), how they have to be accessed from different activities of business processes (call) and finally how these services are provided by the service infrastructure accessing heterogeneous application systems from the ICT layer (binding, provision). Example of service composition. Figure 1-22 shows interplay between conceptual and ICT layers by the example of composing the complex KM service search for experts from the business process layer. On the conceptual layer, this KM service has to be described using knowledge process descriptions and action charts specifying basic input and output parameters. Area of expertise is required as the minimum input parameter. Further input parameters can be specified that describe the context of the situation in which the service is invoked. Examples for context parameters are process, i.e. the business or knowledge process as well as the concrete task that the person is currently engaged in, person, i.e. the profile of the person invoking the service, e.g., areas of expertise or skill levels,

1.6. Knowledge Infrastructures

73

user preferences, e.g., for synchronous versus asynchronous communication channels, products, i.e. electronic resources concerning the area of expertise that have been collected or analyzed by the person, e.g., learning resources, handbooks, reports or lessons learned, applications and devices, e.g., a Web browser on a desktop PC or a mobile application on a smartphone, location, e.g., GPS coordinates or the connection, e.g., wired LAN, wireless LAN or UMTS connection, date and time, normalized according to the time zone, which might help to determine the appropriate way of contacting experts and urgency of the need for an expert.

activity , ti se per s f ex meter o ra rea I: a ext pa t con

activity

activity

KM service layer

e.g., Web services application systems

expert search

concept search

yellow page system

semantic integration system

ts il s er ta xp s de e of ct st nta : li co O ith w

employee search

I: personnel number

O: list of matching authors

author search

I: keywords

O: keywords

keyword search

I: area of expertise

O: list of matching experts

expert search

I: area of expertise

ICT layer

activity

search for experts

complex KM services

basic KM services

final activity

CMS author search employee details content mgmt. system

HR system

check availability

O: current status

initial activity

search for experts

I: employee no., date/time

activity

O: tel.no, email, IM address

conceptual layer

awareness Groupware platform

Figure 1-22. Example of KM service invocation

Execution of the service results in a list of experts, brief descriptions, contact history and information about the (social) relationship to the searcher, e.g., common business acquaintances, contact and availability details, ordered according to the preferences of the experts together with

Service execution

74

Service decomposition

Service binding

Agile technologies

1. Foundation

links to further KM services that can be invoked in order to establish a connection to the selected expert. The middle layer in Figure 1-22 shows the composition of a number of basic KM services into one complex KM service. It also maps the required basic KM services to actual, “real” services offered by diverse application systems that are part of the ICT layer. The complex KM service search for experts is composed of the basic KM services (1) expert search, (2) keyword search, (3) author search, (4) employee search and (5) check availability. The expert search service delivers a list of IDs, e.g., personnel numbers, for experts matching the input parameter of an area of expertise. The author search service requires a list of keywords describing the area of expertise. Thus, the complex KM service search for experts also comprises an integration service for the task of finding keywords that describe the area of expertise, here called keyword search. The keywords are assigned to areas of expertise either in a simple database solution or in a more advanced semantic integration system based on an ontology. With the help of an inference engine, these relationships together with rules in the ontology can be used to determine a list of keywords. The author search service then returns a list of IDs of matching authors or active contributors to the CMS. An employee search service takes the personnel numbers found in the expert search and the author search and returns contact details, e.g., telephone number, email address, instant messaging address. Finally, the check availability service delivers the current status of the experts and a decision on their availability. The ICT layer binds the basic KM services of the conceptual layer to application systems in the current work environment of the searcher that can deliver these services. In the case example, there might be a yellow page system, a semantic integration system, two content management systems, an HR system, a Groupware platform and an instant messaging system that offer Web services fitting to the descriptions of the basic KM services on the conceptual layer. Depending on which systems are accessible to the calling complex KM service, the actual implementation could consist of different service bundles or of all or only parts of the services respectively. Consequently, the description of the complex KM service needs to include some specification of what basic KM services are mandatory and of what combinations of basic KM services are allowed. The KM service infrastructure supports service-oriented, agile or ondemand KM approaches in organizations that take into account decentral developments of KM initiatives. Thus, KM technologies have to operate increasingly on infrastructures that support rapid deployment of relevant tools and systems for ad-hoc, intensive and inter-organizational collaborations. Recently, these dynamic approaches of bringing the right knowledge rapidly to the point where it is needed have been called just-in-time KM, workplace learning or on-demand KM.

1.6. Knowledge Infrastructures

Challenges. When designing and implementing KM infrastructures, KM initiatives can introduce service-orientation as additional guideline. The three-layered KM service infrastructure composes services from heterogeneous applications into specific KM services and supports their discovery, call and provision from activities within business processes. This infrastructure aims at solving the following challenges: Strategic alignment is realized by connecting KM services to the materialization of strategic decisions, e.g., quality leadership, in the form of business processes and corresponding application systems. Deployment of KM services in organizations might profit substantially from both, integration and corresponding alignment with strategic goals. Process orientation is realized by not only focussing on business processes as main drivers for calling KM services, but also on knowledge processes which comprise a procedural view of KM instruments implemented by KM services. Numerous KM measures, procedures, instruments or tools applied in isolation from each other are integrated by bundling complex KM services. Business processes determine which KM services are required in which core business processes, are offered by what service processes, who is responsible for them and what resources are allocated to fulfill them. EKIs consisting of a KM service infrastructure on different levels helps reducing complexity and improving flexibility of KM initiatives. One of the major advantages of EKIs is the ability to build its services once and reuse them frequently. However, the efforts to implement EKIs should not be underestimated. Already established KM services have to be identified and documented. New KM services have to be implemented. KM services have to be composed and decomposed finding appropriate levels of detail. Quality of KM services has to be assessed and documented in order to provide a constant level of quality throughout the knowledge life-cycle. EKIs can be considered as approach of a strategy-based integration of KM services providing a blueprint, i.e. framework and platform for KM services defined in KM initiatives. In the following, the services required for KM initiatives are structured according to an ideal EKI architecture.

75

Strategy

Process

Instrument

Architecture

1.6.3 Architecture of Knowledge Infrastructures This book lays out a comprehensive picture of what tools and technologies are already available to fulfill the vision of an EKI. Goal of this section is the description of an idealized, conceptual layered architecture that structures ICT services to be provided by EKIs. This architecture is then used throughout the book as the structure guiding the presentation of numerous concepts, standards, technologies and tools that are available to implement parts of this architecture. The ideal architecture can be seen as a vision: It poses requirements and asks for services to support knowledge work that EKIs should provide

Ideal architecture

76

1. Foundation

from the perspective of KM, rather than focusing on what current technologies are able to do. They represent an amalgamation of theory-driven, market-oriented and several vendor-specific architectures (Maier 2007). Nevertheless, appropriately combined existent approaches and technologies can provide a good foundation to support effectiveness of information and knowledge work. This book discusses current technologies needed to implement EKIs. This section also presents the vision incorporated by the ideal architecture to what is achievable with today’s technologies and thus gives an overview of the structure of the book. In the architecture, services provided by an EKI can be offered by a central server or by a SOA that connects several servers with the help of a service bus. Services are accessed by participants with the help of (thin) clients. The system integrates all knowledge shared in an organization and includes a variety of services that are either aimed at supporting the participant directly or indirectly by supporting upward layers (Figure 1-23).

Layers

access services authentication, adaptation for devices and applications, participant-, role-, process-, project- and situation-oriented access knowledge services investigation individuation exploration, experience, search, expression, presentation, competence, monitoring expertise meta-data integration data integration

interaction in-form-ation initiation, capturing, networking, manipulation, communication, translation, communities archiving

integration services ontology-based integration

instruction enriching, composition, coaching, assessment

process integration function integration

infrastructure services storage, processing, network, security infrastructure sources organisation-internal organisation-external semi-structured structured semi-structured structured contents from CMS, DMS, data from RDBMS, contents from data from online file server, Groupware data warehouses WWW, newsgroups DBs, GIS Figure 1-23. EKI layers Infrastructure services

The Intranet infrastructure provides basic functionality for storage, processing, synchronous and asynchronous communication, sharing of data and documents as well as management of electronic assets in general and

1.6. Knowledge Infrastructures

of Web content in particular. Sources for structured and semi-structured data can be classified into organization-internal and organization-external sources. Infrastructure services are discussed in chapter 2. Networking technologies enable communication between distant persons and technical devices and are discussed according to network standards along the network layers of the ISO/OSI-model up to application layer where basic network services and protocols are implemented. Infrastructure services comprise services for storage, security, and application infrastructure upon which EKI services of higher layers build. An ontology helps to meaningfully organize and link knowledge elements that come from a variety of sources and are used to analyze the semantics of the organizational knowledge base. Integration services are needed to manage meta-data about knowledge elements and the users that work with the system. Synchronization services export a portion of the knowledge workspace for work offline and (re-)integrate the results of work on knowledge elements that has been done offline. Chapter 3 discusses how integration of data and semantics, functions and processes as well as data about users can be implemented. Data integration is strongly based on the implementation of standards that define character sets, addressing, markup, scopes and schema definitions. Examples are Unicode, XML and XML Schema. A first step to bring semantic meaning on top of data integration is meta-data-based integration. Suitable meta-data schemata have to be selected to structure meta-data which can be serialized as RDF/XML and stored in different types of DBMS from where it can be retrieved using query languages. The next step towards semantic integration is ontology-based integration. Examples for standards described in this section are RDF Schema and OWL. Function integration is necessary to enable remote execution of software and communication between (parts of) applications, e.g., by remote procedure calls or Web services. Process integration through workflows and orchestration of services builds on function integration and aims at coordinating and automating tasks within business processes and of knowledge work. The core knowledge processes search and retrieval, publication, collaboration and learning are each supported by knowledge services. These are key components of the architecture and provide intelligent functions for: Investigation: i.e. services for exploration, search, presentation and monitoring of knowledge elements and experts with the help of e.g., mining, visualization, mapping and navigation tools, Individuation: i.e. services that support appropriation of ideas or results found with the help of investigation services for experience, i.e. reflecting experiences, expression, i.e. externalizing and documenting knowledge, competence, i.e. self-management of competence development, and expertise, i.e. services for discovering, approaching and handling of experts in organizations,

77

Integration services

Knowledge services

78

Access services

EKI - vision and reality

1. Foundation

Interaction: i.e. support of joint creation, sharing and application of knowledge by knowledge providers and seekers with the help of e.g., initiation of interactions by location and awareness management tools, networking with like-minded people, contextualized communication and coordination tools as well as community homespaces for fostering communities of interest, In-form-ation: i.e. services that support capturing, manipulation, transformation and archiving of knowledge elements supported by workflows which are thus turned from more informal artefacts used personally or in interactions into formal organizational knowledge assets, Instruction: i.e. services for enriching material to form learning objects, composing courses by authoring tools, supporting consumption of learning resources by tools for delivering courses, tutoring, coaching and learning paths and services that support assessment of learners. Knowledge services are the core of EKIs and are thought to offer complex KM services for realizing KM instruments. Chapter 4 discusses approaches and technologies to implement these knowledge services. Main aim of personalization services is to provide more effective access to the large numbers of knowledge elements. A portion of the contents and services of EKIs can be organized for specific roles, e.g., with the help of role-oriented portals. Also, contents and services can be personalized with the help of, e.g., interest profiles, personal category nets and personalizable portals. Automated profiling can aid personalization of contents and services. Participants access EKIs either with a thin client, e.g., a Web browser installed on Desktop or mobile devices such as PDAs or smartphones, or indirectly by applications and appliances where the EKI services are integrated into. On the server-side, a variety of services help translating and transforming contents and communication to and from the systems. Functions for authentication and authorization are necessary to protect EKIs against eavesdropping and unauthorized use. Chapter 5 shows how and by which communication channels participants can access knowledge services. They can either be provided in a single point of access, i.e. a portal that enables integrated and personalized access, or by mobile devices and applications for location-independent and thus mobile access. Other services enable platform-independent, manageable access through Web browsers and rich clients or rich thin clients as well as by means of technologies that enable desktop integration and thus are a foundation for personal information management (PIM) and for enabling access according to new metaphors as an alternative to the “traditional” desktop metaphor. This book gives a detailed introduction into the concept of EKI, its layers and building blocks. An ideal EKI that substantially improves knowledge work throughout and beyond the organization is partly a vision that is

1.6. Knowledge Infrastructures

not yet reality in organizations, although there are some innovative companies that have implemented individual productivity environments that come close to this vision. The vision, mainly based on a conceptual view of knowledge work and KM, is reflected in the main structure of the book (levels one or two of the table of contents). But there are many existing classes of technologies and software systems that cover one or more of the layers or building blocks of EKIs and appear in the detailed structure of the book (level three). It is important to understand the relation between functionality typically provided by a certain class of software and the requirements for an EKI building block. Chapter 6 reflects on the potentials of already existing technologies to fulfill the vision and gives an outlook to future developments. So, after completion of this book you will be able to describe the most important technologies used to support knowledge work today, assess their typical functionality and classify them in relation to the conceptual architecture of EKIs.

79

80

1. Foundation

Case Example Two years ago, Fix-IT, a system house recruited Aisha and her colleague, Hans, from the same University and both hold a Bachelor degree with Information Systems major. The two set up a test installation of a new software module promoted by Zip-IT, a big vendor of standard enterprise systems, experimented with it and developed a vision of how it could be applied within the business sector focused by Fix-IT. Also, they identified and connected with internal experts, experts from key customers and from Zip-IT who shared their vision. After convincing Fix-IT’s innovation manager, Aisha became manager of an internal project with the goal of the module’s deployment within her company. She kept with the project until it matured into a product being sold to leading customers. After Hans had left Fix-IT, Aisha was the only remaining person with profound knowledge and long-standing experiences with the module. Consequently, she now is the centre of a network including business consultants, software developers, domain specialists and infrastructure managers of other departments within Fix-IT, Zip-IT and leading customers. She leads a group of seven knowledge workers responsible for operating business application systems according to predefined service levels. Aisha has learned not only to rely on personal communication, but that once the project had left the experimental state, formal documents gradually have grown important. She feels her changed role and the need for approval of even the tiniest change to operative business applications. This particularly includes representatives of customers, but also colleagues, e.g., responsible for infrastructure management, if the change has side effects outside Aisha’s area of responsibility. As Aisha’s knowledge is in high demand, her schedule is consistently filled to capacity. Thus, she has learned quickly how to effectively organize her work. She consequently sets aside time for contemplative work on concepts, but also monitors ongoing developments, collects ideas, reviews proposals and integrates them into the activities within her group. She strongly relies on putting content or messages she receives into agreed documents and communication channels accessible to all people involved in the operation of the product. So, she can reduce face-to-face conversations to a manageable number and efficiently point to digital resources when she is addressed with concrete questions. This translation from the informal, experimental and explorative world of innovation to the formal, routinized and productive world of operative business applications is what she excels in. Aisha does not have any content or documents not part of the EKI which she uses constantly to support all aspects of her work. Her motto is “There is nothing, but experiencing it by yourself, so try and make errors, but if we should translate it into a customer product, it has to go by the rules”.

1.6. Knowledge Infrastructures

Questions and Exercises 1. What are the most important dimensions of knowledge? What opposites

can be distinguished in each of the dimensions? Find examples for knowledge that reflects each of the opposites. 2. Which fundamental perspectives on knowledge can be distinguished? What consequences has that for knowledge management? Describe examples of concrete goals and measures for each of the perspectives. 3. What are the main differences between the “transfer” of data and knowledge? 4. Discuss the relationships between the terms capability, expertise, intellectual capital, knowledge and skill. Find examples for each of these terms. 5. How is knowledge related to organizational strategy and competitive advantages? 6. Sketch Nonaka’s processes of knowledge conversion and find supporting KM instruments for each of the processes. 7. Discuss the definition of the term knowledge management. What lines of development can be distinguished? 8. What goals can an organization aim at when implementing KM? 9. What is knowledge work? Discuss whether or not activities that are part of job descriptions known to you qualify as knowledge work. 10.Discuss whether and to what extent the following positions fulfill the characteristics of knowledge work: air traffic controller, hair dresser, medical doctor, nurse, sales person responsible for selling PCs in a local PC store, service technician for household appliances, e.g., dish washer, microwave. Which types of knowledge are focused in each of these positions? 11.What information and communication technologies could support the implementation of KM instruments? Find examples for one instrument out of each class. 12.Discuss how the term architecture can be used with respect to an organization’s information and communication infrastructure. 13.What are the main advantages of layered architectures? 14.What is a service? Reflect the generic concept as used in business administration and the definition of IT services. 15.What are main components of a SOA? What standards are there to support identifying, describing and invoking services? 16.What is an EKI and how could it support knowledge work?

81

82

1. Foundation

Further Readings There are numerous books and articles on KM and related topics. The following list provides in no way an exhaustive list, but only attracts the reader to some works that provide an overview of many concepts. Alavi, M., Leidner, D. E. (2001): Review: Knowledge Management and Knowledge Management Systems: Conceptual Foundations and Research Issues. In: Management Information Systems Quarterly (MISQ), 25(1), 107-136 A comprehensive overview and discussion of the definitions of KM and KMS that also offers some important implications for further research. Holsapple, C. W. (ed.) (2003): Handbook on Knowledge Management. Vol. 1+2. Springer, Berlin The book draws together a large number of authors and perspectives on KM spanning strategic, organizational, and technological issues and also provides some case studies that show KM in action. Maier, R. (2007): Knowledge Management Systems. Information and Communication Technologies for Knowledge Management. 3rd edition. Springer, Berlin An almost encyclopedic treatise of the facets, concepts and theories that have influenced KM is presented together with the state-of-practice on the basis of a comprehensive empirical study. The book furthermore integrates the concepts, theories and practical findings into a general KM framework consisting of strategy, organization, systems and economics. Mertins, K., Heisig, P., Vorbeck, J. (eds.) (2003): Knowledge Management. Best Practices in Europe. 2nd edition, Springer, Berlin The book comprises practical instruments, techniques and tools that were developed in the course of a number of projects in organizations implementing KM. Nonaka, I., Takeuchi, H. (1995): The Knowledge Creating Company. Oxford University Press, New York A classic in KM and still an interesting read for those that are interested in the characteristics of the approaches implemented in Japanese organizations that helped to create the KM hype in the 90s. Schwartz, D. G. (ed.) (2006): Encyclopedia of Knowledge Management. Idea Group, Hershey, PA. Just like Holsapple’s compilation, the book features a large number of authors and diverse perspectives on KM.

2 Infrastructure

Infrastructure is a term used for a variety of things. The most common contexts for the term in everyday life are roads, water and wastewater, electricity and telecommunication. The basis for an EKI is the network infrastructure. This starts with laying cables into the floors and/or the walls of organization’s buildings or setting up base stations for a wireless network. Today, a medium-sized company building needs several kilometers of cable in order to provide a good network infrastructure for every workplace. Besides the computer network for data communication today there is also still a telephone network for voice communication that forms a separate infrastructure. Although both forms seem to amalgamate in the near future, as current IP telephony and data communication over telephone lines indicate, we will concentrate our considerations on computer networks. Theoretical foundations for computer networks including a layered architecture are presented in section 2.1. Section 2.2 then discusses standard network protocols that are implementations of the layers. Based on the basic network infrastructure, services for data storage, access, messaging and security are offered (section 2.3). More recently, many organizations have also installed an application infrastructure (section 2.3.3) that provides services needed in most enterprise applications (e.g., transaction support).

Overview

On completion of this chapter you should be able to describe the most important network standards for PANs, LANs, MANs and WANs, use a conceptual model that distinguishes several layers within a network that provide different services and build upon each other, identify the most important network protocols for the Internet and categorize them according to layers in the model, distinguish several network devices and explain their usage, describe the most important technologies for storage, access, messaging and security, explain the functionality of application servers and explain how they support enterprise applications.

Learning objectives

84

2. Infrastructure

2.1 Network Infrastructure A classification of computer networks helps to understand the features networks can provide. We will first present a list of several classification criteria before we discuss two criteria, topology and geographical expansion, in detail. The section concludes with illustrating the ISO/OSI reference model, a layered architecture for computer networks. Classification of computer networks

Computer networks can be classified according to a variety of criteria (Table 2-1). The criteria focus either on technical and physical aspects, e.g., the way of transmission, type of media, topology and scale of a network, or on organizational aspects, i.e. who owns and operates the network or who is allowed to access and use the network. Table 2-1.

Network owner

Criteria for classifying networks

criterion

description

example

transmission

mode of transmission

broadcast, point-to-point

topology

basic structure of the physical network

ring, bus, star

scope

geographical expansion of the network

PAN, LAN, WAN

network owner

institution that operates the network

company network, Internet, Value Added Network

user group

group of users that has access to the network

Internet, Intranet, Extranet

function

purpose of the network

front-end network, backend server network, backbone

cost

cost for hardware or lease

50 € per month for ADSL

media

type of physical media

fiber optics, copper cable, infrared light, radio

transmission protocol

structure of messages, message passing, etc.

Ethernet, Token Ring, ATM

performance

network bandwidth

low (< 1 MBit/s), medium, high (> 1 GBit/s)

The distinction regarding the owner of a network is especially important for wide-area networks since LANs are usually owned by the company that also uses the network. The Internet is an open network that consists of a large number of networks operated by different organizations, mostly large telecommunication provider, but also governmental or scien-

2.1. Network Infrastructure

tific organizations. A participant only has to make a contract with one network owner connected to the Internet in order to get access to the whole network. In contrast to that, value-added networks are usually operated by a single organization that connects business partners by providing specialized data exchange services. This is traditionally mostly EDI data, but increasingly more XML data to support business processes. A more technical distinction can be drawn according to the function of the network (part). A front-end network connects clients to the LAN, while the network that connects the servers is referred to as back-end network. This distinction also effects the required speed of the network. While it’s usually sufficient to connect clients with a 100 MB/s connection, the back-end should be faster in order to cope with the traffic resulting from requests from and answers to a large number of clients working with services offered on the servers (e.g., 1 GB/s Ethernet). The term backbone refers to the core network that connects several front- and backend networks with each other and therefore builds the central component in a network. It is often organized in a ring topology and has the highest demand for fast components so that 10 GB/s Ethernet is desirable. Cost and performance are usually closely related both in terms of leasing cost for access to public networks and in terms of hardware cost for network components to build a private network. The faster the network should be, the higher the cost. Two examples should point that out. While a 2 MB/s leased line is around 250 € per months a 8 MB/s line is more than four times more expensive. Similar to that, simple 24 port switches that allow to connect 100BaseT Ethernet devices are available starting around 100 €. Enterprise switches with 24 Gigabit ports and high switching performance cost around 2000 €. Other criteria shown in Table 2-1 will be discussed in conjunction with network standards in section 2.2 and infrastructure services in section 2.3.

85

Network function

Cost and performance

2.1.1 Topologies Machines linked in a network communicate by sending messages through a communication medium. In analogy to the traditional mail system these are called packets. The actual content of the message is wrapped in transport data about sender and receiver as well as data to control the transport process. In this section, alternatives to structure communication networks are explained with regard to physical connections between machines. A topology represents the physical communication connections (edges) between machines (nodes) in a network. The architecture of a network is fundamentally determined by the mode of transmission which can be broadcasting of messages or point-to-point communication. In broadcast networks, all nodes are connected to the same physical medium. Messages sent by one machine are received by all others, although they can either be addressed to one single communication

Broadcast networks

86

Point-to-point networks

Bus network

Ring network

2. Infrastructure

partner (unicast) or to all nodes in the network (broadcast). Some networks allow addressing of messages to a subset of the network which is termed multicasting. A main challenge in broadcast networks is how to utilize the network medium most efficiently which largely depends on management of media access. In point-to-point networks (sometimes also called peer-to-peer networks), only pairs of machines are connected through a physical medium. Distant communication partners thus regularly will not be connected directly to each other. In this case, a message needs to be transmitted via intermediate stations that receive and forward the message to its destination. A challenge in this class of networks is how to find the best out of multiple alternative routes that connect source and destination of the message. In contrast to broadcast networks, point-to-point networks are only capable of unicasting. Topologies of broadcast networks. Bus and ring networks are two basic types of topologies of broadcast networks which are often combined in practice, e.g., a ring network connects multiple bus networks. The left hand side of Figure 2-1 shows a bus network. All nodes are connected to a shared communication medium, e.g., a copper cable, that passively transmits messages in both directions. No routing or forwarding of messages is necessary because every connected station receives all messages. In a ring network, every node is connected to exactly two other nodes. A message is passed only in one direction from one node to the other until its destination is reached. If one station breaks down, the whole communication is blocked. To enhance the reliability of the network, redundant communication channels can be established that allow bypassing defect machines. The nodes usually regenerate the physical network signal with the effect that network size is not limited with respect to the number of participating machines, but by length of the connections between two nodes.

bus network Figure 2-1.

Topologies of broadcast networks

ring network

2.1. Network Infrastructure

Topologies of point-to-point networks. Four basic architectures of pointto-point networks can be distinguished (Figure 2-2): In a star network, all nodes are physically connected to a central node that handles the entire communication between all nodes. The advantage of this topology is that maintenance and control of the network are simple, because it can concentrate on the central node. A major disadvantage is the dependence of the whole network on performance and availability of the central node. In a tree network, communication between two nodes always runs over hierarchically superordinated nodes. This architecture thus can be seen as to be composed out of interconnected star networks. Network control is performed by superordinated nodes. Thus, the tree network has the same advantages and disadvantages as the star network: easy administration, but dependence on central nodes. In a mesh network, every node is connected to two or more other nodes. If all nodes are directly connected to each other, we speak of completely intermeshed networks. Mesh networks are very reliable and can grow without central control. A disadvantage of this architecture is that it can be complex to find the best route between source and destination. In a loop network, every node is connected to exactly two other nodes. Every single node controls network traffic. Thus, network control is more complex than in the case of centralized network architectures such as star or tree networks. In contrast to a ring network, communication between nodes can happen in both directions and communication media not necessarily have to be of the same type. The nodes in a loop network play a more active role than in a ring network as they forward messages and not just refresh physical signals.

star network Figure 2-2.

tree network Topologies of point-to-point networks

mesh network

87

Star network

Tree network

Mesh network

Loop network

loop network

88

2. Infrastructure

2.1.2 Geographical Expansion Networks can also be classified according to their scope, i.e. the geographical area they cover. Table 2-2 gives an overview of commonly distinguished classes of networks with respect to their geographical scope. Table 2-2.

Network classification according to scale

interprocessor location examples, distance network for ... 1 m workplace 10 m conference room 100 m company building

network class personal area network (PAN) local area network (LAN)

1 km university campus 10 km city 100 km country 1000 km continent 10.000 km planet Personal area network (PAN)

Body area network (BAN) Local area network (LAN)

Metropolitan area network (MAN)

Wide area network (WAN)

metropolitan area network (MAN) wide area network (WAN) the Internet

Personal area networks (PANs) connect devices of a single person. Often, the term implies the use of some form of wireless technology to connect e.g., personal digital assistants (PDA), notebooks and cellular phones, or to connect a PC to a printer and a scanner. The term body area network (BAN) is sometimes used to denote very short ranging personal area networks between components of a wearable computer, e.g., computer system that is integrated into a jacket. Local area networks (LANs) are used to share hardware and software resources within a workgroup or an organization. They are restricted in size and usually span a single building, site or campus. Although geographic expansion of small LANs and wide reaching PANs may overlap, there is a clear distinction as LANs connect several computers belonging to different people, e.g., in a work group whereas PANs connect several devices belonging to one person. A metropolitan area network (MAN) covers a larger area within a city or region. The term is used in two contexts. One denotation refers to the interconnection o separate LANs of a single organization via leased lines to build a virtual LAN that covers all buildings of the organization on a campus or within a city. The second denotation refers to a network that provides centralized services like Internet access (e.g., with WLAN access points) or cable television to private homes. A recent development is highspeed wireless Internet access via MANs (IEEE 802.16 standard, UMTS). A wide area network (WAN) enables communication with very high bandwidth (e.g., 10 GBit/s) over large distances, e.g., between states,

2.1. Network Infrastructure

countries or continents. It basically consists of hosts, e.g., servers, switches, routers, and a communication subnetwork that connects them. The network usually is owned and operated by a telephone company, an Internet service provider or by public authorities. Challenges in WANs are bandwidth management, cost accounting, scalability and high reliability. Last, but not least the Internet is the network of interconnected networks spanning all continents of our planet (section 2.3.2, 126ff).

89

Internet

2.1.3 Layered Network Architecture Computer networks can be seen as hierarchical systems with several layers. Each layer provides certain services to the higher layers which leads to an increasing degree of abstraction for higher layers. The higher layer can therefore be recognized as a service consumer whereas the layer below can be seen as a service provider. The specification of the service is the interface between both layers. There exist a number of layered architectures of which the ISO/OSI reference model is the most important one from a theoretical perspective. The Open Systems Interconnection (OSI) model has been defined by the International Standards Organization (ISO). It specifies seven layers that help to understand network systems (Figure 2-3). Application Layer

7

Presentation Layer

6

Session Layer

5

Transport Layer

4

Network Layer

3

Data Link Layer

2

Physical Layer

1

Application Protocol Presentation Protocol

6

Session Protocol

5

Transport Protocol Network Data Link Physical

3

Internal Protocols

4 3

2

2

1

1

transmission medium

Data Station A

Figure 2-3.

7

Protocol

3

Protocol

2

Protocol

1

transmission medium

Mediator A

Mediator B

Data Station B

ISO/OSI reference model

The physical communication in this model is top-down from the highest to the lowest layer on the sender side, then horizontal over the network media and then bottom-up from the lowest to the highest layer on the receiver side. From a conceptual point of view, each layer on sender side communicates with the layer on the same level on receiver side.

ISO/OSI reference model

90

2. Infrastructure

Example

Imagine the owner of a small company in Wellington, New Zealand, who recently met a business man from Munich, Germany, on an industry event in the USA. When she comes home from the event, she wants to send him a message, but she does not know his address details. All she knows is the name of the man and the company he works for. (a) So she gives her secretary the message she wants to send and the instruction to transfer it as soon as possible. (b) The secretary looks up the address details, puts the message into an envelope and puts the address on the envelope. She is not sure about the correct value of stamps she has to put on the envelope, so she gives the letter to the post office of the company. (c) In the post office, they look up the rate and put stamps on the envelope. Then, they hand it over to a logistics provider that handles all the letters and parcels for the company. They do not know on which way the letter will travel to Germany, nor do they care. (d) The logistics provider handles that on its own. The letter goes by truck to the central post office and from there to Auckland airport. The plane with the letter leaves to Singapore, which is a central trade center for passengers as well as goods and flies to Frankfurt/Main airport in Germany. The message moves on to Munich by train until it reaches the post office of the company at its destination. (e) There, the clerk reads the address on the envelope and brings it to the secretary of the receiver (f) who removes the envelope and finally (g) hands it over to the business man. The company owner back in New Zealand thinks that she has communicated directly with the business man in Germany. In fact, she has only communicated directly with her own secretary. The same system applies to the ISO/OSI reference model. Each layer has a distinct purpose that will briefly be described in the following section. The application layer (7) gets data that has to be communicated from the application. The presentation layer (6) is concerned with syntax and semantics of transmitted data. It uses high order data structures (e.g., bank account records). The session layer (5) handles session information and time-outs, so that user-specific information is only saved as long as applications need it. The transport layer (4) provides a virtual channel for communication that can either be connection-oriented or connection-less. The former means data can rely on data passed on earlier. The latter means data is transmitted according to fire-and-forget mechanisms. Connection-oriented transportation can be thought of as having the telephone system as role model. A user picks up the phone, dials a number, communicates with the remote person and hangs up. Connectionless transportation on the other hand can be thought of as having the postal system as role model. Each message/letter carries the full address and is routed through the system independently. The advantage of connectionless transportation is fault tolerance, whereas connection-oriented transportation allows for billing and a guaranteed level of quality (quality of service, QoS). The network

ISO/OSI layers

2.2. Network Standards

layer (3) is responsible for identification of communication partners which is implemented via unambiguous identification addresses. The data link layer (2) takes care of reliable reconstruction of the transmitted signal by adding checksums to the data. Finally, the physical layer (1) provides access to the transmission medium by modulating the signal. In between two data stations, there can be several mediators that refresh the signal and route the network packets. They may use different protocols to communicate with each other. Real end-to-end communication is only established on layers four and above.

91

Mediators

2.2 Network Standards The following sections discuss concrete implementations for different layers of the ISO/OSI model. These implementations are standardized network protocols. Afterwards, we also present network hardware which can be categorized according to the ISO/OSI layers they are based on.

2.2.1 Physical and Data Link Layer Starting from the bottom, transmission media which can be wired or wireless have to be examined. For wired media, fibre channel and copper cables can be distinguished with copper cables being further divided into twisted pair, coaxial and power cable. The different types of copper cables vary in cross-section and shielding, which are the main characteristics that influence attenuation and liability to interference. Today, copper is mainly used for end-user connectivity. Most long range cables are fiber optical because of better transmission characteristics. With more and more fiber optical cables produced, costs decrease so that this material becomes affordable and is also used for connecting servers in order to satisfy increasing bandwidth demands. This leads to a further increase in production and decreasing costs so that it seems only a matter of time until fiber optical cables are used to connect end-user computers (classification according to function in Table 2-2 on page 88). Wireless media can be divided into radio connections and optical connections that differ not only in wavelength of the signal (1-5 GHz for radio vs. 3-300 THz for optical connections), but also in their diffusion model. Where optical methods usually use point-to-point connections, radio frequency emitting devices usually send in all directions (broadcast). Both types have two important sub-types: terrestrial radio transmission and satellite transmission for radio connections as well as infrared light and directed laser for optical connections.

Media, cable, fiber, copper

Wireless media, radio, optical

92

2. Infrastructure

Table 2-3 gives an overview of the network standards discussed in the next section with respect to geographical expansion of networks explained in section 2.1.2, 88ff. Table 2-3.

USB

Firewire

IrDA

Overview of network standards cable-bound

wireless

PAN

USB, Firewire

IrDA, Bluetooth

LAN

Ethernet, Token Ring

WLAN, DECT

MAN and WAN

ATM, FDDI, X.25, FrameRelay, Sonet/SDH

GSM, GPRS, EDGE, HSCSD, UMTS, UWB

Protocols for personal area networks. There are two important standards each for wired and wireless connections between devices in personal area networks. The purpose of these standards is to provide easy connectivity, low implementation costs and small yet robust physical interfaces. For wired connections, USB (universal serial bus) is the dominating standard and can be seen as a successor of the serial RS/232 interface. It is used to connect peripheral devices, e.g., pointing devices and keyboards, as well as digital cameras, scanners and printers to computers. It can also be used to connect peripheral devices with each other (USB2go), although this is still rarely the case. USB version 1.1 allows transmission with up to around 1 MBit per second. Version 2.0 is downwards compatible and specifies transfer speeds up to 480 MBit per second. Version 3.0 further speeds up data transfer rates to 4.8 GB/s by introducing additional connection pins and using an additional optical connection while keeping backwards compatibility. First devices implementing the new standard are expected to be available in late 2008. A competing standard for USB 2.0 is the IEEE specification 1394, commonly known as iLink (Sony) or Firewire (other companies). It is mainly used for connecting external high-speed devices like DVD burners or hard disks to computers (especially by Apple Computer, Inc.) and in the video industry for connecting digital video camcorders to computers. The standard can be used for transfer speeds up to 400 MBit per second (1394a). A successor with doubled transfer speed has been specified as 1394b. Wireless connections in PANs are still mainly realized via infrared light signals based on the IrDA standard (Infrared Data Association). Most devices available only support version 1.0 that provides transfer rates of 115 kBit per second. There are also two faster versions that provide 4 MBit per second (version 1.1, also called Fast IrDA) and up to 16 MBit/s (version 1.2, sometimes called Very Fast Infrared, VFIR). However, both are rarely used. All versions have a signal range of 1-2 meters. Mobile phones are the largest group of devices that heavily rely on IrDA for wireless connection to computers and especially notebooks. Optical connec-

2.2. Network Standards

tions rely on a direct line of sight between both communication partners. This leads to easy interruptions in the communication, e.g., by shakes that move or rotate the mobile phone just a little so that connection to the computer gets lost. Bluetooth, a radio standard for PANs, becomes more and more adopted, because radio transmissions are much easier to handle. It is designed to facilitate any kind of wireless connections, e.g., connect keyboards and mice to computers, connect headsets to mobile phones, or connect digital cameras to printers in order to directly print photos. Bluetooth was specified by the Bluetooth Special Interest Group (BSIG) initiated in 1998 by Ericsson, IBM, Intel, Nokia and Toshiba. Version 1 was designed to connect devices within a 10-100 meters range with 1 MBit/s transfer rate and operates at 800 mW transmission power in the 2.4 GHz frequency band. It is a technique for ad-hoc connection of devices where one device declares itself as master and up to seven other devices can connect as slaves and form together a so-called Piconet. Devices can also take part in more than one Piconet simultaneously. The resulting overlapping Piconets form a socalled Scatternet. To support such different deployment scenarios as described above, every device has to support a number of profiles. A profile specifies a concrete set of protocols that span one or more layers. The profiles are divided into basic and advanced profiles (Table 2-4). Table 2-4.

Examples for Bluetooth profiles

type

profile

typical purpose (connect ...)

basic

generic access

discovery of remote devices and services

service discovery

discovery of supported profiles

dial-up networking

notebook to mobile phone (Web access)

LAN access

PDA to WLAN access point (PPP)

generic object exchange

calendar items between PDAs (OBEX)

synchronization

PDA to PC

advanced basic imaging

scanner or digital camera to PC

hands free

headset to mobile phone

hardcopy cable replacement

PC to printer (formatted text and images)

human interface device

mouse and keyboard to PC

local positioning

GPS device to PDA

Despite its increasing dissemination, there are still some interoperability problems between devices of different vendors that are the result of inaccurate specification or sloppy implementation. Therefore, testing

93

Bluetooth

94

Bluetooth 2.0 EDR, Ultra wideband, NFC

Ethernet, IEEE 802.3

CSMA/CD

Token Ring, IEEE 802.5

2. Infrastructure

whether two devices interoperate is principally advisable, even if both devices support the same profile. Two recent protocol standards are trying to overcome some problems of Bluetooth. The ECMA and ISO standard ultra wide band (UWB) is designed to significantly speed up data transfer rates of Bluetooth to up to 480 MB/s for a distance of 3m by using a large frequency range (3,1-10,6 GHz) and 110 MB/s for 10m. In order to avoid interference with other radio standards in parts of the frequency bands used, the transmitting power is very low (0.6 mW). The transmission method is called orthogonal frequency-division multiplexing (OFDM). This technology will be integrated into Bluetooth 3.0 which will follow up the current Bluetooth 2.0+EDR (enhanced data rates) with a speed of 3 MB/s (gross, 2.2 MB/s net). The other problem of Bluetooth is the robustness of the pairing mechanism of devices. Near-field communication (NFC) has been adopted as Bluetooth 2.1+EDR to overcome this problem and provide a fast and reliable pairing mechanism for two devices within a very short range ( in text and avoid that they are interpreted as commands, these characters are replaced by special codes. Another application for entities is abbreviation, e.g., for large URLs.

< for Ü for Ü

comment

Comments that are ignored by the processor. They begin with .

processing The otherwise purely declarative elements of XML instruction also contain a procedural element, the processing instruction (PI). PIs are written in a special tag with the form . PIs are ignored by the XML processor, but must be passed through to the application that executes all PIs it knows.

CDATA

0) y=x; ]]>

CDATA sections are arbitrary strings not interpreted by an XML parser. They can contain further markup that is not interpreted or provide information for other interpreters, e.g., JavaScript code. CDATA sections begin with .

3.1. Data Integration

An XML document consists of the prolog and the instance of the document. The prolog is an obligatory element that states the XML version, the type of the document and other characteristics of the document. It has two components, the XML declaration and the document type declaration. The XML declaration states the XML version and can contain further elements: . This example document uses version 1.0 of the XML standard (attribute version), has to be interpreted in UTF-8, a subset of Unicode (attribute encoding) and is independent of other documents (attribute standalone). Formally, the prolog is a processing instruction. The document type declaration sets the rules for the structure of XML documents. It is either part of the prolog or the prolog points to it. The most popular format for document type declarations is XML Schema. It offers the advantage that the schema code conforms to the XML standard. It has more options to describe the document, a large number of predefined data types and allows for complex, user-defined data structures and data types together with mechanisms for inheritance, primary key attributes and foreign key attributes for reference purposes (section 3.1.4) The document instance contains the elements, attributes, entities, processing instructions, comments and strings (CDATA) that form the document. XML documents can be mapped to a tree, with each XML element represented by a node. Listing 3-1 shows an example XML document. Authors of our text book Ronald Maier 1968 +43-512-507 38000 Thomas Haedrich 1977 [email protected] Rene Peinl 1975 [email protected] Listing 3-1.

Example XML document

159 Prolog

XML declaration

Document type declaration

Document instance

160

3. Integration Services

Description of example

The document contains a prolog with XML version and a specification of the character set used. The document instance consists of a root element, also called document element, . The team’s name is an attribute to the element. contains the elements and . The latter element in turn contains an id as attribute and four elements that comprise the first and last name, email address or phone number and the year in which the person was born. Figure 3-3 shows a portion of the example XML document mapped to a tree. team

descr

person

name=“EKI author team“

person

id=“3470“

Authors of our text book

firstname

lastname

Ronald

Maier

Figure 3-3. Well-formed XML documents

year-of-birth 1968

phone

+43-512-507 38000

Example for XML document represented as tree

An XML document must comply to the following set of rules for the use of elements, attributes and entities in order to be well-formed: there is only one root or document element containing all other elements, opening tags must be matched with corresponding closing tags. Note that XML is case-sensitive. Exception: empty tags draw together opening and closing tag in case there is no content between the two, elements must be nested properly, so if element is nested in element , then comes before , attribute names have to be unique within one element, so an element cannot contain more than one attribute with the same name, attribute values must be surrounded by quotes (" or '), entities are declared before they are used. Namespaces. XML markup defines a vocabulary, also called a markup vocabulary, in Figure 3-3 for example a vocabulary to describe persons that work together in a team. Once a markup vocabulary is defined, it can be reused by other XML documents. XML documents in turn might contain elements from multiple markup vocabularies. This could create problems of recognition and collision. Software processing XML documents has to recognize tags and attributes it should process. Also, collisions between element or attribute names from different sources might occur.

3.1. Data Integration

These considerations require that document constructs have universal names the scope of which extends beyond their containing document. XML namespaces provide a mechanism for creating universally unique names for elements and attributes. An XML namespace is a collection of names, identified by a URI which are used in XML documents as element types and attribute names. Using namespaces requires that every name in the XML document specifies the namespace in which it is defined, e.g., . Thus, names from XML namespaces appear as qualified names which contain a single colon separating namespace prefix and local part. The prefix selects a namespace and is mapped to a URI. The prefix, in this case book, is an abbreviation for the actual namespace assigned in a namespace declaration. title denotes the local part of the qualified name. The prefix must be stated in a namespace declaration attribute in either the start-tag of the element where the prefix is used or in an ancestor element, i.e. an element in whose content the prefixed markup occurs. Exceptions are the two namespaces xml and xmlns. The prefix xml by definition is bound to http://www.w3.org/XML/1998/namespace. The prefix xmlns is used only for namespace bindings and is not itself bound to any namespace name. The namespace declaration applies to the element where it is specified and to all elements within the content of that element, unless overridden by another namespace declaration, e.g.:

161

Qualified names

Namespace declaration

Namespace scoping

The Adventures of Huckleberry Finn

In this case, the namespace declaration applies to the element book and consequently to the element title that is surrounded by the element “book”. This is also called namespace scoping. A default namespace applies to the element where it is declared (if that element has no namespace prefix), and to all elements with no prefix within the content of that element. A default namespace is declared by leaving out the prefix in the declaration: The Adventures of Huckleberry Finn 0140620648

In this case, the default namespace for the element book is urn:publicid:books and the same applies for the un-prefixed element title whereas the number is taken from the prefixed namespace isbn.

Namespace defaulting

162

3. Integration Services

3.1.4 XML Schema

Purpose of XML Schema

Local vs. global definition

Elements and subelements

Complex and simple types

We have already discussed the notion of a well-formed XML document which requires that the document conforms to a set of syntactic rules. As XML is used for semi-structured data, there is also the need to define more constraints for documents. For this reason, a number of schema definition languages have been developed that serve this need, e.g., Document-Type Definition (DTD), RELAX NG, Schematron or XML Schema. The purpose of a schema is to define a class of XML documents and thus it is not surprising that the term instance document is often used to describe an XML document that conforms to a particular schema (=class). However, neither instances nor schemes need to exist as documents, e.g., they may be stored as fields in a database record, but when referring to “documents” constructed for a specific schema, this is mostly called schemes and instances as if they were documents or files respectively. XML Schema does for XML what a table definition in SQL does for relational databases. Therefore, it is an integral component of native XML databases (See “Storage in an XML database” on page 185). XML schemes can be designed in two different ways: with a focus on types or with a focus on the structure of the document. When types are focused, all type definitions reside on the same level in the XML tree of the schema file and the structure of the whole XML document that is modeled emerges from composing the elements and their complex types. This is called global definition and provides maximum reuse as complex types can be referenced at multiple points in the schema. When structure is focused, then the type of each element is directly specified within this element instead of referencing it with the type attribute. This is called local definition. As reuse of type definitions is not possible when using local definitions, this is only suitable for small schemes. In order to explain the various elements of XML Schema, we refer to the example XML document instance in Listing 3-1. The instance document consists of a main element, team, and the subelements descr and person. The latter in turn contains other subelements, e.g., firstname, year-of-birth, that contain strings or numbers rather than subelements. Some elements have attributes, in this case team and person. Elements that contain subelements or have attributes are complex types, whereas elements that contain numbers, strings etc., but do not have subelements are simple types. Attributes always are simple types. Complex types and the simple type year are defined in the schema for persons in teams. The other simple types are built-in simple types defined in XML Schema. These are similar to data types in databases. Examples are boolean, integer, decimal, float, double, date, time. Additionally, some data types are specific to XML Schema, such as ID, a unique identifier attribute type, IDREF, a reference to an ID, entity, language or name. Listing 3-2 shows the corresponding XML schema with global definitions:

3.1. Data Integration

163

Listing 3-2.

Example XML schema “team”

The schema consists of a schema element and several subelements, particularly element, attribute, complexType, and simpleType that constrain the appearance of elements in instance documents. Each of the elements in the schema has a prefix xsd: associated with the XML Schema namespace through the declaration in the schema element: xmlns:xsd="http:...". The prefix xsd: is used by convention to denote the XML Schema namespace. The prefix xsd: is also found in the qualified names of built-in simple types, e.g., xsd:string which clarifies that these types belong to the vocabulary of the XML Schema language and not to the vocabulary of the schema author. Definitions of complex types typically contain a set of element declarations, element references, and attribute declarations. For example, teamtype is defined as complex type comprising two element and one

Prefix xsd:

Complex types

164

minOccurs and maxOccurs

Required, optional, prohibited Deriving simple types

Restrictions

Content models

Choice, sequence and all

3. Integration Services

attribute declarations. Consequently, elements in instance documents of type teamtype must consist of two elements descr and person and can contain an optional attribute name. The elements must appear in the order they are declared. The descr element is declared as the built-in type string, therefore descr elements in instance documents must be strings. The definition comprises a complex type persontype also defined in the schema. Finally, the teamtype definition contains the attribute declaration name which assigns the type string. As attributes cannot contain other elements or attributes, all attribute declarations have simple types. Listing 3-2 is an example for a schema with global definitions. The attributes minOccurs and maxOccurs restrict the number of elements allowed. In our case, their has to be at least one person described as part of a teamtype and the maximum number is not restricted. The attribute minOccurs can take on values between 0, meaning that the specified element is optional, and any positive integer number. maxOccurs can take on integer numbers starting from 0 or the term unbounded meaning that the number of elements is not constrained. Attributes may appear once or not at all. This can be declared with the use attribute to indicate whether they are required, in our case id, optional, in our case name, or prohibited meaning that the attribute must not occur. Listing 3-2 declares several elements and attributes that have simple types, some built-in to XML Schema, e.g., string and decimal, and the simple type year that is derived from the built-in type integer. Both, built-in simple types and their derivations can be used in all element and attribute declarations. In our case, the derived type year is used in the element definition year-of-birth. New simple types are derived from existing simple types, i.e. restricting the range of values allowed for the new type as a subset of the existing type’s range of values. The restriction element is used to indicate the base type which can be a built-in or an already derived type, and to identify the facets constraining the range of values. In our case, year is restricted by the two facets minInclusive and maxInclusive to values between 1900 and 2100. XML Schema offers a number of facets to restrict allowed values that are quite similar to the options found in high-level programming languages. Examples are length restrictions, patterns, enumerations, and inclusive and exclusive minimal and maximal values. The definitions of complex types in Listing 3-2 declare sequences of elements that can or must appear in the instance document. The occurrence of individual elements may be constrained depending on the values of minOccurs and maxOccurs. XML Schema also provides constraints that apply to groups of elements appearing in a content model. The choice group element allows only one of its children to appear in an instance. In Listing 3-2, the complex type persontype contains a choice group with the two elements email and phone. Hence, in an

3.1. Data Integration

instance document, a person element must contain either an email element or a phone element. The choice group and the declarations of the elements firstname, lastname and year-of-birth are children of a sequence group. The sequence group assures that the elements appear in exactly the declared order. Finally, the all group specifies that all elements in the group may appear once or not at all, but they may appear in any order. The all group is limited to the top-level of any content model. Summing up, XML Schema provides a powerful instrument to define structured data sets and offers even more mechanisms not touched here. One important difference to data definitions as found in databases is the use of mixed content, i.e. XML elements mixed with plain text in document instances. To enable character data to appear between the child elements of team, an additional attribute, called mixed on the type definition of team is set to true. Listing 3-3 gives an example instance document with tagged elements conforming to the schema in Listing 3-2.

165

Mixed content

This document contains a list of Authors of our text book . Ronald Maier is born in 1968 . His office phone number is +43-512-50738000, but he can also be reached by email. ... Listing 3-3.

Semi-structured text as example for mixed content

3.1.5 XSL Transformation In order to make semi-structured data interchangeable, it is not sufficient that both, the sending and the receiving application support XML. At first glance, there is no difference between two vendor-specific XML file formats and two file formats in a proprietary binary format. In both cases, one file cannot be read in the other program and vice versa. Typically, it is costly and cumbersome to write a converter for binary formats and usually the structure of the format is not publicly available. For XML data, there is a standardized way to translate one format into the other. The language that supports conversion is called extensible stylesheet language (XSL). XSL consists of the two parts transformation (XSLT) and formatting (XSL-FO, formatting objects) and is based on the XPath language used to address parts of XML documents. The name of XSL suggests that it fulfills a similar purpose for XML documents as cascading stylesheets (CSS) fulfill for HTML documents. However, CSS can be used for XML documents as well as for HTML documents to specify display parameters whereas XSL is specifically designed for converting XML documents

XSL parts

166

3. Integration Services

from one structure (or schema specification) to another. We will first discuss XPath, before having a closer look at XSLT and an outlook to XPath version 2 and XQuery, a query language for XML documents similar to SQL for relational databases. XSL-FO will not be further discussed as it does not support transformation of documents, but is mainly intended for formatting XML documents for PDF or other print output. XPath. XPath is used to address elements and attributes in an XML document. The syntax is similar to specifying a path in a file system. The forward slash (/) is used as delimiter and also specifies the root node of the XML tree. The XML document in Listing 3-4 is used as reference to illustrate XPath and XSLT features. The result of an XPath “query” is always a node set containing zero, one or more XML nodes (elements or attributes). XPath instructions consider the hierarchical tree structure of XML documents. This is necessary because a descr element as child of team can mean something different than the same element as child of person. Authors of our text book University Ronald Maier professor production industry Rene Peinl project manager IT consulting Thomas Haedrich consultant Listing 3-4.

XML file for XSLT transformation

3.1. Data Integration

Paths can either be absolute (starting with the root node) or relative (starting with the current context node) with absolute paths always starting with a forward slash. The XPath expression /team/descr is an absolute path whereas person/firstname denotes a relative path. This relative path is only valid if the team node is the current context node for the XPath instruction. It is also possible to address nodes irrespective of their location in the XML tree. Two slashes (//) are used to express this. Therefore //descr selects all description elements regardless whether they are children of team or of person. It is also possible to use wildcards for selecting several nodes with different names. The XPath expression person/* matches all child elements of person. One can further create unions over node sets selected by different XPath instructions with the "|" operator. The XPath expression //firstname | //lastname selects all firstname and lastname elements anywhere in the document. Formally, an XPath statement consists of three parts, an axis, a node test and optionally a predicate axisname::nodetest[predicate]. An example is child::person[@id='3472']. The axis is used to specify further context used to evaluate the node test and determines whether the node should be an element or an attribute and whether it is above, beneath or at the same level in the tree. A node test is used as part of an expression in order to retrieve one or more nodes within the context set by the axis, in our case person elements that are children of the current context. The predicate is used to restrict the node test, in our case to the person identified by id=3472. Table 3-2 summarizes available axes and their function and lists some abbreviations that can be used instead of writing the full axis name. Additionally, examples of XPath expressions are given and described in terms of their meaning with respect to Listing 3-4 (starting with node organization, id="UIBK"). Abbreviations are given in parentheses, e.g., @ for attribute. The default axis is child. Table 3-2.

167 Absolute vs. relative paths

Formal XPath constituent

XPath axis types and examples

axis

example

description

ancestor

ancestor::*/descr

selects description elements of the team the organization is in

ancestor-or-self

ancestor-orself::*/@id

selects the id attributes of organization and team

attribute (@)

attribute::id

selects the id attribute of organization

child

child::descr

selects the descr element of organization

descendant

descendant::/@id

selects the id attributes of all persons at this organization

168

3. Integration Services Table 3-2.

XPath axis types and examples

axis

example

description

descendant-or-self (//) descendant-orself::*/@id

selects the id attributes of organization and of all persons at this organization

following

following::*/descr

selects the descr elements of the next organization and all persons there

following-sibling

following-sibling::*/descr

selects the descr elements of all following elements at the same level in the hierarchy

namespace

namespace::*

selects all elements defined in the same namespace

parent (..)

parent::*/descr

selects the descr element of team

preceding

preceding::*/descr

selects the descr elements of all elements before the organization element

preceding-sibling

preceding-sibling::*/descr

selects the descr elements of all preceding elements at the same level in the hierarchy

self

self::descr

selects the descr element of the current organization

A predicate can be used to further restrict the selection to a subset of the nodes selected by the axis and node test. A predicate can be any XPath expression that is valid when being evaluated on the unrestricted result set as the context. Restrictions can be made regarding characteristics of the result set, e.g., position()=2, or regarding the value of a child element or attribute, e.g., @id='3472'). Expressions with calculations can also be used as restrictions, e.g., @id mod 2 = 0. In calculations, the basic mathematical operators can be used (+, -, *). The division operator is div and there is also a modulo operator mod. The Boolean operators and and or can be used as well as = and != (inequality) operators and the relational operators >, = and = 50000). The OPTIONAL operator deals with data that might be missing. It returns the requested attributes if present, but returns also the other results that have all the required attributes, but not the optional one. The FROM keyword known from SQL references a URL for retrieval of the RDF data. SPARQL also supports ORDER BY clauses like SQL, allows to return only a specified number of results using the LIMIT and OFFSET operator and is able to return only unique results using the DISTINCT operator. Furthermore, SPARQL distinguishes four query forms. SELECT: returns all or a subset of the variables bound in a query pattern match. CONSTRUCT: returns an RDF graph described by substituting variables in a set of triple templates. ASK: returns a boolean number indicating whether a query pattern matches or not. DESCRIBE: returns an RDF graph that describes the resources found.

3.2. Meta-data-based Integration

191

3.2.4 Meta-data Dimensions One challenge for content management is to have a leading system for every kind of resources that holds the most current, elaborate and correct version (“single source of truth”). It controls or even reduces redundancy by identifying copies and different versions of an instance of the content and introduces a synchronization process as well as controlling mechanisms that prevent changes in decentral content versions without permission of the content owner. These contents represent important business objects of an organization and are often used as meta-data for documents as well. This meta-data is more valuable, the richer the used data type is. The simplest form of metadata representation is the use of basic data types (text, number). A first step towards richer information is to reference a list or table in the same IS that stores the documents where additional information about the metadata is provided that can be accessed by clicking on the meta-data value of a document. The richest kind of information is the reference to the information object in an organization-wide information system that is the leading system for the respective type of information and provides relevant actions in conjunction with the information type, so that users can click, e.g., on the author of a document and are redirected to the organization's yellow pages where they can find additional information about the person like department, skills or past projects. In this section we discuss the most broadly usable meta-data dimensions answering the generic questions who (person), where (location), when (date, time) and why (organization and process) that can be used in order to structure meta-data used to integrate resources throughout an organization in an EKI. Person dimension. Today, most office workers have to access several enterprise systems and commonly they have different login information for each of them. Many company security policies regulate that passwords have to be changed regularly. Furthermore, passwords need to satisfy complexity criteria, e.g., they must be at least 8 characters long, contain upper and lower case letters as well as characters and numbers. That leads to users often writing down passwords so that they do not have to remember them which certainly is not desirable. Additionally, time is wasted for login processes and user data is spread across applications although some user preferences, e.g., colors or fonts, are applicable to any system. User integration denotes efforts to integrate data about users that is managed by different applications. Advanced knowledge services require integration of data about users across applications in order to, e.g., personalize services and contents. We will discuss the basic terms involved in that topic in the following paragraphs before we describe account management comprising wide-spread techniques for managing user data. Finally,

Enriching meta-data

User integration

192

3. Integration Services

identity management is discussed as an umbrella term for sophisticated techniques that provide more advanced possibilities to handle user data. Identification, authentication, authorization

Privilege, role, security group

User account, master and transaction data

Centralized access to PCs, files and printers

LDAP directory services

Terms and definition. In order to gain access to an enterprise system, a user has to have a (user) account to log into the system. One component of an account is the data that uniquely identifies a user, e.g., a user name, a personnel number or an ID stored on a chip card (identification). The second component is data that is used to verify that the person is really the user he or she pretends to be, e.g., a password, an image of the fingerprint or a certificate stored on a chip card (authentication). The third component is data used to control access to certain functions of the system or data stored in the system (authorization). The latter are called privileges. A privilege grants access to a set of functions or data. Every user could theoretically be assigned a personal set of privileges. As this is time-consuming, users can be combined to groups and privileges are assigned to those groups instead of single users. Alternatively, privileges can be combined to a set of associated privileges called role and users are assigned to one or more of these roles. Although both approaches are different and could be described as top-down and bottom-up, they usually lead to similar results. The role-based approach is often favored because the term role is also used in organizational contexts to describe a set of tasks a user has to fulfill, e.g., project manager or work group manager. Users often hold several roles simultaneously, particularly with respect to systems where many technical roles are defined on a finer level of detail, so that one organizational or process role resolves to several technical roles. Together with account data, there are two other sets of user data that are stored in an enterprise system. One is the user master data and contains, e.g., the user's real name, address, telephone number. The other is data about user preferences and transaction data like past user actions the system uses to customize its look & feel (section 5.1.2, 370ff). Both together (sometimes even the second one alone) build a user profile. Account management. A first approach to centralize management of user data mainly targets user accounts on OS level in order to unify access to central resources like printers and files on file servers. Users should also have the possibility to log into different computers using their centrally managed account. That has greatly simplified the management of user accounts as the user's login information does not have to be managed locally on every computer the user accesses, but only once. The central application that holds data about user accounts is often an LDAP directory service (section 2.3.1, 114ff). Users and resources are both assigned to a hierarchical structure that is used to partition administration tasks. The hierarchical structure is inherited from the LDAP hierarchy, but is usually structured according to DNS domains. The DNS names have a structure that fits to the LDAP objects (Listing 3-17).

3.2. Meta-data-based Integration

193

peinlpc.eki.virtual-corp.de c=de,o=virtual-corp,ou=eki,computer=peinlpc [email protected] c=de,o=virtual-corp,ou=eki,cn=peinl Listing 3-17. Example for LDAP storage of DNS names

Similar to the notation in email addresses, a user name is specified using the @ symbol. The hierarchy specifies the object classes country (c), organization (o), organizational unit (ou) and common name (cn). Please note that the object class computer is not specified in the LDAP standard, but in a vendor-specific schema extension. Therefore, it can also be called server or machine depending on the software. Centrally managed accounts are called domain accounts (referencing DNS) in distinction to local accounts on a PC. Besides the DNS hierarchy used to group user accounts and resources, there are also security groups that are independent of domains and can be used to, e.g., grant access to a file share to a project team consisting of employees of various organizational units. Employees are assigned to the same security group and access is managed based on the group instead of single users. In some directory service implementations, there are a number of group types that fulfill different purposes and can be nested. Windows 2000 for example distinguishes the group types security group and distribution group (or email messaging and both types can be either domain-local, i.e. users from any domain can access resources from one domain, domain-global, i.e. users from one domain can access resources in any domain or universal, i.e. users from any domain access resources from any domain. Permissions to access a specific resource are finally managed in socalled access control lists (ACLs). ACLs consist of one or more entries that map user accounts or security groups to permissions. Possible permissions are: list folder content (folder only), read, execute, write (create new files), change (override existing files) and full access. Each permission can be explicitly granted or denied, otherwise it is unspecified. Access is only granted if explicitly stated in the ACL, otherwise denied. The reason for this is that permissions can be inherited from objects higher in the hierarchy (folders or domains) which can lead to conflicts due to users belonging to multiple groups with potentially contrary privileges. Denying permissions always overrides granting permissions. Inheritance of permissions together with group memberships complicates evaluating a concrete user request to read or write a file. So there is a trade-off between security and convenience of administration on the one hand and fast access and lightweight implementations on the other hand. Example implementations

Security group

Access control list

194

3. Integration Services

for account management are Windows Active Directory or OpenLDAP together with a Samba server that provides similar functionality. Identity Management. Account management is well established in medium-sized and large companies, but it represents only a first step to integration of data on users since it is limited to access to OS and network drives. However, access to enterprise applications should also be centrally managed. The term identity management has been coined for a holistic approach to do that. A user can have many different accounts that are used to access various applications in the network and thus can have multiple so-called digital identities. Applications can identify these accounts by attributes of the user, e.g., email address, personnel number or user name. Definition of digital identity

A digital identity is the set of attributes that uniquely identifies a physical user throughout all applications on the network.

Components of identity management

Figure 3-6 shows the components that form an identity management solution. The components can be grouped into layers that reach from the storage of user data to the concrete services the user gets from the solution. application

lifecycle

security

data repository

Figure 3-6. Data repository layer

Security layer

single sign-on

personalization

provisioning

authentication

LDAP directory

delegated administration

longevity

authorization

auditing

database

Components of an identity management solution

Like account management, identity management solutions usually build on LDAP directories to store user data, but DBMSs could be used as well (section 2.3.1, 114ff), especially if they have a virtual LDAP interface, that allows to add and retrieve data over LDAP. On top of the data repository, the security layer provides methods for authentication and authorization (section 2.3.2, 126ff). Furthermore an auditing component logs all changes and thus makes them traceable. Besides different protocols, e.g., Kerberos, CHAP, EAP, and authentication methods, e.g., user name and password, biometric, smartcards, the important aspect for identity management is that not all installed enterprise

3.2. Meta-data-based Integration

applications provide LDAP authentication. Therefore, wrapper applications have to be developed which are a kind of middleware and translate between central and application-specific user accounts, so that central authentication is sufficient for accessing the application. The life-cycle layer consists of a provisioning and a longevity component. Provisioning is the automation of procedures and tools to manage the life-cycle of an identity: (1) creating an identifier for the identity; (2) linking it to authentication providers; (3) setting and changing attributes and privileges; and (4) decommissioning the identity (De Clercq, Rouault 2004). Propagation of data from a central repository to applications that do not support central user management is crucial and has to be developed on a case by case basis. Longevity tools create historical records of identities. These tools allow examination of the evolution of identities over time. Longevity is linked to the concept of attestation or the ability to attest what actors had access to what resources in what time frame, irrespective of whether they actually exercised their access rights, which is a matter of auditing. Life-cycle and security layer together provide the core components of identity management in a narrow sense. The components on the application layer bring the benefits of identity management to the user. Based on the core components, users get single sign-on functionality, so that they only have to log in once to get access to all applications they need. It is further possible to personalize applications based on one consistent set of user data (section 5.1.2, 370ff). Delegated administration is a means to decentralize administrative tasks by giving administration privileges for defined parts of the infrastructure, e.g., a user group or a server, to a number of employees. Self service goes one step further and allows end-users to change their own password, address information and other person-related data on their own. The additional benefit is that the central management overhead is kept small and there is no danger of bottlenecks for changing user data or other administrative tasks. Identity management targets not only account data, but also other userspecific data that resides in enterprise applications. The usage of user profiles and data about user actions across applications can significantly improve the perceived usefulness of system functions and contents if systematically used for personalization. Software vendors that provide identity management solutions include IBM, Microsoft, Novell, Oracle, SAP and Sun. The solutions consist of multiple parts. For example, Oracle bundles its LDAP directory service Internet Directory together with Certificate Authority, Application Server Single Sign-On, Directory Integration, Directory Provisioning Integration Service, and Delegated Administration Services. Microsoft builds on Active Directory (Application Mode) for its Identity Lifecycle Manager product that consists of Identity Integration Server and Certificate Lifecy-

195

Life-cycle layer

Application layer

Delegated administration, self service

Personalization

Product examples

196

Identity management and Internet

Identity Federation

3. Integration Services

cle Manager and is often accompanied by third party products like Omada Identity Manager for provisioning workflows. In September 2008, Sun released their Identity management solution as open source software, consisting of OpenSSO Enterprise, Java System Directory Server and Java System Identity Manager. There are even efforts to expand identity management to the Internet. Microsoft has taken a first step with its .NET Passport initiative. A .NET passport can be used to get access to a number of Web sites from Microsoft Network (MSN), hotmail and developer network (MSDN) to several other e-commerce sites. A monopoly for managing digital identities might harm privacy. Several major software vendors have founded the Liberty Alliance under lead of Sun Microsystems that offers its own identity management solution on the Internet based on a federated infrastructure, i.e. several servers with trusted relationships between them. There is also an open source approach called OpenID. Where OpenID is an approach for the Internet, Identity Federation works in Extranet scenarios, where user repositories of multiple organizations work together to provide a homogeneous zone for authentication. Location dimension. Use of geo data has gained a lot of attention since the release of Google Earth in 2005, although it was preceded by Nasa World Wind in 2004. These applications exceed traditional route planning software in the capability to provide high quality satellite images in addition to the vector-based road data and allow for overlaying additional information, e.g., points of interest (POI), using XML-based file formats like KML (Keyhole Markup Language). Even more use is made of their Web-based counterparts like Google maps, Yahoo maps or Microsoft maps that allow for mashup applications displaying e.g., photos, real estate offerings or crime rates. We will explain the basic mechanisms underlying these applications in the next section, before discussing exchange formats for geo data and finally introducing a hierarchy of geo-codable objects relevant in organizational contexts.

Coordinate reference systems

Geo data: terms, storage and querying. Geo data can be defined as data about objects, forms of terrain and infrastructure at the surface of the earth that bears a spatial reference. They consist of geometric, topological, graphical and attribute data. Geometric data allows defining location and forming by referencing a coordinate system and specifying x-, y- and zcoordinates or longitude, latitude and elevation. Topographic data highlights height differences by displaying isolines. Graphical data provide a visual impression of how the area represented by the map looks in reality, e.g., by satellite images. Attribute data represents domain-specific information, e.g., income distribution in an area. There are several different coordinate referencing systems like the World Geodetic System (WGS 84) currently used by the Global Position-

3.2. Meta-data-based Integration

ing System (GPS), the International Terrestrial Reference System (ITRS) maintained by the IERS or the Universal Transverse Mercator (UTM). Therefore, the need for transformation between different systems arises, especially between the geographic (geodetic) systems referencing an ellipsoid (WGS, ITRS) and the Euclidean projected systems (UTM). There are also local spatial coordinate systems with limited reach, like the German DHDN90 (Deutsches Hauptdreiecksnetz 1990), which still play a significant role since they can be more precise than world-wide reference systems with typical tolerances of one meter. The coordinate system is referenced using a spatial reference identifier (SRID). Geometric data can basically be divided according to its dimensionality into zero- (point), one- (line), two- (shape) and three-dimensional data (corpus). There is a growing number of relational DBMS that support storage of geo data in a specialized data type. Examples are Oracle (since version 8i), PostGIS, which is an extension of the open source DBMS PostgreSQL, and Microsoft SQL Server 2008. These systems do not differentiate between geo data of different dimensionality and either provide one general geo data type or two different ones for geographic (ellipsoid) and geometric (flat) data. In addition to the above mentioned lines and shapes, they allow storage of multi-lines and different specialized shapes like circles, arcs and rectangles. Geo data can be differentiated in vector data and raster data. Vector data defines geometric forms like the ones described above and is stored as reference to a coordinate system whereas raster data is stored as pixel data with reference to lines and columns and a color value. Where vector data is well suited for binary distinctions like road or forest, coast or water, raster data allows to see gradual changes that are reflected in minor color differences. Both kinds of data can be visualized within the same map, e.g., to display territorial borders on a satellite image. The Open Geospatial Consortium (OGC) defines a number of standards for geo data, most notably the Simple Features Specification (SFS), which exists in versions for CORBA, OLE/COM and SQL. The SFS for SQL 15 is widely implemented in version 1.1.0 (current in version 1.2) and defines a Well-Known Text (WKT) or Well-Known Binary (WKB) representation for geo data as well as a number of functions that operate on them. In order to interact with system-specific internal data types for geo data, OCG SFS specifies from and to operators that convert from and to WKT and WKB representations. In addition to that, the above mentioned DBMSs also provide from and to operators for GML and/or KML. The operators centroid (center), distance, area, length and boundary determine certain aspects of a given shape and IsEmpty, IsValid,

15

URL: http://www.opengeospatial.org/standards/sfs, last access 2008-10-06

197

Storage

Vector vs. raster data

Geo-data standards

Geo operations

198

3. Integration Services IsClosed, IsSimple, IsRing (closed & simple) return true or false

depending on the shape. The relation between two geo shapes can be determined using the operators equals, disjoint, union, intersection, difference, touches, crosses, contains, within and relate. The relate operator returns the relationship between two shapes in form of a matrix, the so called DE-9IM (dimensionally extended nine-intersection matrix). The columns stand for border, interior and exterior of shape B, the lines for those of shape A. The example in Figure 3-7 shows that A lies completely in the exterior of B (represented by the ones in column e) and vice versa (represented by the ones in line e). The one in cell Ab:Bb show that A and B have a touching border, i.e. a common point. B

A touches B

Figure 3-7. Exchange, GML and KML

GeoRSS

b

i

e

b

1

0

1

A i

0

0

1

e

1

1

1

9 insertion matrix

Example for a 9 insertion matrix

Exchange of geo data is mainly handled in XML formats like GML (Geographic Markup Language) or KML (Keyhole Markup Language). The former was specified by the OCG16 and is a vendor-independent standard whereas the latter is a common format specified by Keyhole, Inc., now owned by Google and made popular as the standard for Google Earth. GML was constructed in correspondence to the OCG SFS and is primarily designed for 2D geo data, although a third dimension can be specified. The term simple in SFS refers to geometry and allows the description of points, lines and polygons as well as homogeneous (multi-point, multi-line or multi-polygon) or heterogeneous collections (mixed points, lines and polygons). One of the design goals has been exchange of not only geo data, but also of other describing properties of geographic objects. To do that, one has to define a GML schema. KML additionally provides a wealth of styling attributes that allow controlling line, icon and label styles. Ground overlays can be used to draw images upon the map, e.g., for displaying crime rates or flood simulations. An extension of RSS called GeoRSS is recently gaining attention since it can be reused in existing RSS applications without geo capabilities and still provides the most important features for overlays in geo applications. 16

URL: http://www.opengeospatial.org/standards/gml, last access 2008-10-06

3.2. Meta-data-based Integration

Its usage grows rapidly, because Google, Microsoft and Yahoo support it on the one hand and RSS is widely known on the other hand. Despite these advantages, there is also the drawback, that there are multiple substandards. Using the W3C Basic Geo Vocabulary17 which specifies a point as and is outdated, but still many usage examples for the popular mapping services are based on it. The favored alternative is called simple geo encoding18 which specifies point, line, polygon and box elements, as well as relations to other Web resources with a relationship tag, e.g., “resides-in”, “is-located-at”, a featuretype tag, e.g., “plant”, “sales office”, a featurename, e.g., “Innsbruck headquarters”, “Munich office”, and floor, elev(ation) and radius. All discussed data exchange formats can be integrated into the map display at the browser level using JavaScript. At first, the reference to the JavaScript API has to be established. After that the API calls can be used to load and display geo data, as well as set additional parameters that cannot be controlled from within the data file, like size in pixels of the map display, map center, zoom level and display of navigation controls. Geo coding denotes translation of postal addresses into geo coordinates. Accordingly, reverse geo coding is the translation of geo coordinates into human readable addresses. Geo coding is the basis for routing from one point to a second one. The primary actions for locations are “drive to” or “drive from” in order to get driving directions. For routing, it is necessary for the system to know the roads that can be used by the desired transport vehicle. After that, established shortest path algorithms can be used to find the best route. In order to consider different road types and user preferences, travel cost can be assigned to roads. This technique can also be used to consider traffic jams by temporarily assigning high costs to the affected road sections. Modern navigation solutions also provide additional geo data overlays called points of interest (POIs). These POIs can be, e.g., fueling stations, sights or speed limits. This data can be used to answer user queries like “show the way to the closest hospital” or “determine the route along the most interesting sights”. In a business scenario, the POIs can be used to show customers their closest sales office. To do this, the location of the customer has to be determined, which can be done using IP geocoding. Since all public IP addresses are registered to a person or organization including their location, a rough geo location can be looked up19. Precision is around 25 km (with large deviations depending on the region, e.g., urban or rural). In an organization, there are a number of entities that have a geo location starting from single workplaces and reaching up to sales regions like EMEA (Europe, Middle East and Africa). In between are rooms, e.g., 17 18 19

URL: http://www.w3.org/2003/01/geo, last access 2008-10-06 URL: http://georss.org/simple, last access 2008-10-06 URL: http://whatismyipaddress.com/, last access 2008-10-06

199

Actions

Hierarchy

200

3. Integration Services

meeting rooms, floors, e.g., floor plans, buildings, subsidiaries, e.g., a company campus, states, countries and continents that can be of interest. Fine granular levels are suited for finding single employees, planning office layout and optimizing space usage and communication whereas intermediate levels are suited for getting driving instructions to the subsidiaries. Coarse-grained levels can be used for reviewing sales data or planning new subsidiaries. IS that are affected therefore also differ. At fine levels, the primary IS would be a telephone book application or collaboration server. Floor plans usually are available as PDF file and not directly integrated with geo applications. Different buildings on a campus can be displayed directly within the geo application and all big vendors currently invest in 3D capabilities so that organizations can model their buildings and display them within a geo application. For driving directions, it is common practice to replace formerly dominating PDF files with online route planning services integrated into the organization’s Web site. At the highest level, business intelligence systems are affected which are mostly not yet geo-enabled, but can be configured to display data as an overlay to images such as maps. Storage

Exchange

Date and time dimension. Date and time data types have been established in IS for a long time. The SQL-92 standard defines date, time, timestamp as well as interval as basic data types and therefore every DBMS adhering to SQL-92 supports them. There are also some basic operations defined for calculating the difference between two datetimes or adding a specified interval, e.g., two days, to a datetime value. Conversions to and from strings that contain datetime data in different formats are usually supported. Still, there is no support for higher level semantics like days of week, calendar weeks, public holidays or leap years. Even time zones and daylight saving are optional and not included in every DBMS. Input of dates is sometimes limited to a given format, e.g., mm/dd/yyyy, whereas other formats like dd.mm.yyyy are not supported. Therefore, it is not possible to query a database for entries that were created on Mondays if there is only a creation date stored. Main source for higher level date and time data is a collaboration server that takes care of these issues. For exchanging calendar data, iCalendar (RFC 2445), or briefly iCal20, is still the prevailing format, although it is pure text and not XML-based. It specifies the four main components event, to-do, journal entry and free-busy-lookup that can be accompanied by time zone and alarm components to include further details. It is similar to the vCard standard (see 4.3.3, 292ff). iCalendar was specified in 1998 by Microsoft and Lotus and most collaboration systems support it by now. Nevertheless, XML-based formats have some advantages and therefore most large ven20

Not to be mixed up with Apple’s calendaring software.

3.2. Meta-data-based Integration

dors have established own XML-based formats that replace iCalendar and are used internally as parameters for Web service calls. Listing 3-18 gives an example for a calendar event, represented in the Microsoft Exchange 2007 XML language occurring on 28th of February 2008 from 7 to 9.30 p.m. US east coast time (GMT-8) for a workout at the gym. The date/time dimension has the specialty that there are two parallel dimensions. On the one hand, days can be seen as days of month which in turn can be aggregated to quarters, half-years and years. On the other hand, days are also weekdays that can be used as a division for years. Additional complexity arises from the difference between calendar years and fiscal years that is given in many organizations.

201

Hierarchy

2008-02-28T19:00:00-08:00 2008-02-28T21:30:00-08:00 OOF false 00000 Exercise Gym false false false true Listing 3-18. Example for XML calendar event in MS Exchange 200721

The basic data type time span can represent an appointment on a higher semantic level or even more specifically a business meeting. Timestamps that are used as meta-data for a document can be ordered along the lifecycle of the document (see chapter 4.4, 305ff). Timestamp values for creation, last modification, last access, due date, retention time and destruction are stored by default in most document management systems. Some basic actions are related to time data like “convert to my timezone” and “calculate difference to today”. Interactions with users’ personal calendars build another category of helpful actions. Examples are “show time span in personal calendar” or “reserve in calendar”. Organization and process dimension. Organization and process are two separate dimensions, but since they are closely interrelated on the one 21

See also URL: us/library/aa563496(EXCHG.80).aspx.

http://msdn2.microsoft.com/en-

Semantics of time

Actions

202

Hierarchy

Primary IS for organizational data

Roles

Exchange

Visualization

Actions

Hierarchy

3. Integration Services

hand and there is not much established support in EKIs on the other hand, they are presented here underneath a single headline. The primary dimension of organizational data is often hierarchical and characterized by abbreviations, synonyms and frequent changes through renaming or reorganization. Different versions of an organizational unit (OU) have to be mapped to a single entity in order to get a semantic representation. An OU has a type which is closely related to its level within the hierarchy of OUs, e.g., division, department, group or competence centre. Nevertheless, there can be different types of OUs on the same level. The leading system for organizational data is usually the respective module of an ERP system, e.g., SAP OM, which is often not directly accessible for all employees. However, the hierarchy itself, without any further information may also be stored in a number of other systems, especially the central user directory which is often structured according to the OU hierarchy and has an own object class for OUs. The link between organization and person dimension is drawn with roles. Especially the manager of an OU is a relevant role, since this kind of data can be used for approval or review processes. Similar to that, information about the role of a team or OU assistant can be used in processes for delegation and request of support services. There is no wide-spread standard for exchanging data about organizational structures. Maybe this is due to the fact that organigrams are often considered confidential in order to prevent competitors from hiring employees in important positions. Therefore, exchange of data about the organization structure is often limited to exchanging graphical representations as PDF documents or vector graphics such as Visio drawings. Organigrams are the most familiar representation means for organization data. Their notation is simple. Even in current standard office software, there is built-in functionality to draw and embed organigrams in office documents. They can be embedded in Web pages by using, e.g., SVG or VML (Vector Markup Language, Internet Explorer only). Desirable actions for OUs that serve as meta-data for documents are, e.g., “get head/assistant of OU”, “get employees in OU”, “get sub/super OU”, “get topics/responsibilities of OU”. Even a basic action that links to the respective organigram would be helpful. Process. The process dimension could also be called activity dimension in a broader sense, since activities within business processes are as interesting as activities outside business processes, e.g., in projects. According to Nonaka’s model of the hypertext organization, there are three parallel systems of activities: business system layer: routine tasks are organized in value chains, (sub)processes, functions and partly supported by workflows, project team layer: projects can be subdivided into milestones, which can again be refined into tasks,

3.2. Meta-data-based Integration

knowledge base layer: activities in communities and knowledge networks support developing topics and competencies in organizations. Development of interchange formats for process models has gained momentum by business process management middleware. Despite its name it represents an enhancement of application server and Web service composition software and still struggles to bridge the gap to business processes. BPEL is currently the most wide-spread XML language for process models (see section 3.4.2). However, there is a lot of criticism to BPEL and employees of business departments often prefer working with tools that use less technically constrained languages like BPMN (Business Process Modelling Notation), although models in BPMN are not directly executable in contrast to BPEL models and have no associated XML representation. Mappings to BPML (Business Process Modelling Language), XPDL (XML Process Definition Language) or BPEL are required for exchanging data. The ARIS Toolset and ARIS Business Architect are common modelling environments, especially in the German-speaking countries, due to their integration with SAP Netweaver. Those specify an own XML language for event-driven process chains called ARIS modelling language (AML). BPEL offers a broad implementation base in business process management suites and XPDL can represent all aspects of BPMN models including layout. Using UML diagrams to model business processes does not seem to prevail over BPMN, XPDL and BPEL (section 3.4.3, 231ff). For visualization, event-driven process chains (EPK) and BPMN play an important role. To maintain overview in complex processes, a hierarchical structure is also advisable for graphical representations. The ARIS Publisher for example offers process models as Java applets that allow OLAP-like drill-down and roll-up operations to change the level of detail within a process from value chains (which have a different notation) to processes, subprocesses and workflows. The highest level overview of processes in an organization is called process landscape. Some workflow management systems on the market are still working with different notations that may be based on petri nets, or proprietary notations. For project management, project plans and their visualizations in Gantt charts are commonly used, e.g., with rich clients like Microsoft Project and TaskJuggler or Web-based tools like Oracle Projects and Zoho Projects. Apart from tools for social network analysis, there are no specific management tools aiming at the knowledge base layer. Typical actions for process (steps) that act as meta-data for documents are, e.g., “get previous/following process”, “get all documents associated with this process”, “get superior process” and “get process owner”. Typical actions within a social network analysis application are “get directly connected persons”, “get persons connected over n hops”, “get network centres” or “get isolated persons”.

203

Exchange

Visualization

Actions

204

3. Integration Services

Summary. Figure 3-8 gives an overview of all steps in this section, shows a layered perspective on meta-data, accompanies the overview given in Figure 3-4 at the beginning of the chapter and already relates to the next section about ontology-based integration. The lowest layer shows options for storing meta-data. A number of languages can be used to query this meta-data that are related to the type of meta-data store on the one hand, e.g., SQL is used for querying meta-data in a relational database, but can also be chosen independent of the metadata repository since many DBMS provide more than one query language, e.g., most recent RDBMS support XPath, some also XQuery. The result set can be represented in multiple forms, but for meta-data the use of RDF/XML is suggested. example statement

store best practice

person reason

structure / harmonize

type

location

RDF

topic time

TEI

+

sematical

RQL, RDQL, RDFQL, SerQL, SPARQL, ...

structural

SQL, OQL, XPath/XQuery, free text

store

metadata

XML-DBMS

Figure 3-8.

process

Ontology based on OWL

DOI PRISM LOM Meta-data schema based on RDF-Schema Dublin Core MPEG-7

represent / serialize query

find learning object for skill

find expert for document

RDBMS

OODBMS

document

metadata

document

search index

Layered view on meta-data

The semantic structure of the results should conform to a common meta-data schema, which could either be industry-specific or comprehensively usable. It can also be used to enhance query capabilities. Nevertheless, using a meta-data schema is only a first step towards semantically describing resources. In order to answer more complex queries, an ontology is needed which provides additional mechanisms to describe relations between types of resources.

3.3. Ontology-based Integration

205

3.3 Ontology-based Integration In addition to meta-data integration, ontology-based integration provides standards and technologies to integrate knowledge elements from different systems on the conceptual level. Thus, it is not descriptions of data or Web resources alone that is brokered from system to system, but “understanding” of what data means, its semantics and context. Many of these standards and technologies build on XML. Figure 3-9 shows a four-step model leading from unintegrated to semantically integrated (text) documents.

text documents and data base records (pre-XML) Figure 3-9.

XML documents for single domain using single vocabularies

XML taxonomies and documents with mixed vocabularies

XML-based ontologies and automated reasoning

From unintegrated text documents to semantic integration (Daconta et al. 2003, 3f)

The first step does not take advantage of XML as the standard for structuring text documents which the second step introduces. The third step relies on the definition of namespaces and XML Schema as a meta-language to define and limit the allowed structure and contents of documents. This requires the use of (semantic) meta-data standards for the description of documents or, more generally, Web resources (section 3.2). The fourth and final step heads towards machine-understandable semantics and automated reasoning about documents and is called semantic integration in this book. Knowledge modeling techniques and methods differ with respect to the degree of formality. On the one hand, methods and techniques from the fields of AI and knowledge-based systems are highly formal and represent knowledge in the form of ontologies or domain models that can be processed by computers (section 3.3.1). On the other hand, knowledge mapping techniques are less formal and often primarily serve as a tool for human beings to better understand the (highly aggregated) structure of important areas of knowledge or competence and their relationships to, e.g., persons, groups or other organizational units that create, hold, seek, distribute or apply knowledge (section 4.1, 242ff).

From data integration to semantic integration

206

3. Integration Services

Semantic Web stack

Figure 3-10 gives an overview of the Semantic Web stack (Berners-Lee et al. 2001) covering both, data and semantic integration. Based on XML, a number of standards have been developed that provide means to make statements about Web resources and to relate Web resources to each other. The W3C standard RDF was discussed in section 3.2.2. On the next level, the Semantic Web stack provides a language that supports the design of vocabularies, e.g., classes for instance RDF specifications, called the RDF Vocabulary Description Language or, shortly, RDF Schema (section 3.3.2). Whereas RDF and RDF Schema have been around for some time and can be considered as fairly stable language specifications, there is still a lot of debate going on about what elements should constitute higher levels of the Semantic Web stack. There seems to be some convergence on the level of ontology with the Web Ontology Language (OWL, section 3.3.3). However, standards for specifying entire logic frameworks, exchanging proofs and thus building trust between agents still remain to be seen. Concepts for these ambitious aims are discussed in section 3.3.4.

semantic integration logic framework, proof, trust Web Ontology Language (OWL) RDF Vocabulary Description Language (RDF Schema) Resource Description Framework Syntax Specification (RDF/XML) data integration XML Schema Extensible Markup Language (XML) Uniform Resource Identifier (URI)

Extensible Stylesheet Language Transformations (XSLT) Namespaces Unicode

Figure 3-10. Standards of the Semantic Web stack

3.3.1 Ontology Knowledge modeling aims at formal description of (documented) organizational knowledge that can be processed by computers, e.g., the Knowledge Interchange Format (KIF), and at visualization of topics that are of inter-

3.3. Ontology-based Integration

est in a KM initiative and/or that are supported by contents of an EKI and their relationships. There are relationships (1) between topics and persons, (2) between topics and ICT systems, especially which documents and other resources contain information on a certain topic as well as how they are related to each other and (3) relationships between topics themselves. Several groups of authors have extended methods, techniques and tools that were originally developed to model knowledge used in knowledgebased systems. Examples are the CommonKADS method (Schreiber et al. 1999) or the tools OntoEdit and OntoBroker (Staab et al. 2001). The two terms ontology and taxonomy are used widely for the results of modeling efforts. Depending on the semantic richness of the constructs that can be used to formalize topics, knowledge objects and their relationships, some authors distinguish between (simpler) taxonomies and (more powerful) ontologies. In the following, these two terms and their usage with respect to EKIs are briefly reviewed. The term taxonomy denotes the classification of information entities in the form of a hierarchy, according to the presumed relationships of the real-world entities that they represent (Daconta et al. 2003, 146). A taxonomy can contain definitions and explanations, synonyms, homonyms and antonyms, as in a thesaurus. A taxonomy is often modeled as a hierarchy of terms and can be used as the semantic basis for searching and visualizing a domain, e.g., a collection of documents. Ontologies in KM are formal models of an application domain that help to exchange and share knowledge with the help of ICT systems. “An ontology is a (1) formal, (2) explicit specification of a (3) shared (4) conceptualization” (Gruber 1993, 199). (1) An ontology has to be formal which requires that the ontology is machine-readable. However, there are different degrees of formality of ontologies, from a repository of common-sense knowledge contributed by ordinary people like Open Mind Common Sense22 over a thesaurus like WordNet23 onto ontologies capturing formal theories provided by domain experts like Cyc24. (2) Explicit specification means that concepts and relationships as well as constraints on the use of concepts are defined openly and not left to the interpretation of the ontology’s users. (3) Shared refers to the requirement that conceptualizations made in an ontology have to be agreed upon by a group of people that intend to use the ontology for knowledge exchange. (4) Finally, conceptualization is an abstract model, a

22 23 24

URL: http://openmind.media.mit.edu/, last access: 2008-11-01 URL: http://wordnet.princeton.edu, last access: 2008-11-01 URL: http://www.cyc.com, last access: 2008-11-01

207

Foundation in AI

Taxonomy

Definition of ontology

208

Elements of ontologies

Taxonomy and ontology compared

3. Integration Services

representation of a domain or phenomenon which investigates concepts of that domain or phenomenon that are relevant to the ontology’s users. Ontologies are therefore developed to provide machine-processable semantics of information sources that are accepted by a group of users and facilitate knowledge sharing and reuse. Ontologies are not static, but evolve over time. An ontology not only defines basic terms and relations comprising the vocabulary of a topic area, but also comprises rules for combining terms and relations to define extensions to the vocabulary. Ontologies model (1) objects in domains, (2) relationships among those objects, (3) properties, functions and processes involving the objects and (4) constraints on and rules about objects (Daconta et al. 2003, 190). Thus, ontologies support clear-cut, concise, semantically rich and unambiguous communication between persons aided by EKI and/or between different (parts of an) EKI. Compared to the term taxonomy, the term ontology is usually used not only to describe definitions of terms and basic relationships between terms, e.g., is_a-relationship, but also to support an extended set of relationships, e.g., symmetric, transitive or inverse relationships, and reasoning about concepts that are defined in the ontologies. More specifically, ontologies provide the concept of rule that is used e.g., to check not only syntactic, but also semantic validity of a statement or that is used to derive new relationships between terms from existing ones. Semantic rules, e.g., in the form of inference rules, describe how knowledge can be gained from existing statements. An example is: if two companies operate in the same industry and the same geographic region, then they are competitors (Listing 3-19). FORALL Company1,Company2,Sector1,Region1 is_Competitor(Company1,Company2) >Region1:Region; "operates_in_sector"->>Sector1:Sector] and Company2:Company[ "operates_in_region"->>Region1:Region; "operates_in_sector"->>Sector1:Sector]. Listing 3-19. Example for semantic rule in F-Logic

Types of ontologies

The definition of ontology covers different types of ontologies that play a number of roles in EKI development (Fensel 2004, 5f): domain ontologies capture knowledge of a particular type of domain and are thus restricted to the context of this domain, meta-data ontologies provide a vocabulary used to describe contents in an EKI, e.g., the Dublin Core meta-data standard,

3.3. Ontology-based Integration

common-sense ontologies capture basic notions and concepts for e.g., time, space, state, event and relationship that are valid across several domains, representational ontologies comprise definitions of ways to represent knowledge and are not restricted to particular domains, e.g., frame ontology defining concepts such as frame, slot, slot constraint that can be used to explicate knowledge in frames, method and task ontologies provide concepts specific to particular problem-solving methods, e.g., the concept correct state in a proposeand-revise method ontology, or concepts specific for particular tasks, e.g., the concept hypothesis in a diagnosis task ontology. Ontologies can be formalized with the help of a number of languages, e.g., F-Logic, that are in turn supported by tools, e.g., Ontobroker25. However, the term ontology is sometimes used to describe conceptualizations on a spectrum that extends from weak to strong semantics starting from taxonomy, via thesaurus and conceptual model to logical theories that describe semantically rich, complex, consistent and meaningful knowledge (Daconta et al. 2003, 156ff). In KM, ontologies facilitate communication, search, storage and representation of knowledge (O’Leary 1998, 58). Methodically, ontology modeling methods build upon object-oriented modeling methods extended so that they cannot only be used to develop software during build-time, but also can be used as explicit element of the user interface during run-time. Most organizations that are about to implement or have implemented EKIs have also created at least a minimal taxonomy or ontology (O’Leary 1998, 58). However, the development and continuous maintenance of an ontology requires a substantial amount of effort. Also, ontologies developed individually in organizations are likely to be incompatible and thus cannot be used to share knowledge across organizational boundaries. Consequently, there is a need for standardization, both in the language used to develop an ontology and also with respect to the concepts of ontologies which is pursued by the world ontology projects cited above. As the portion of documents stored in XML formats and the number of tools that can export and import XML documents increase, it is not surprising that XML has been proposed as basis for describing contents of documents and relationships, for a standardized taxonomy of topics and ultimately, for comprehensive, machine-processable standardized ontologies. Examples are RDF Schema (section 3.3.2) and OWL (section 3.3.3).

25

URL: http://www.ontoprise.com, last access: 2008-11-31

209

Formalization

Applications in KM

210

3. Integration Services

3.3.2 RDF Schema RDF vocabulary description

RDF Schema and object-orientation

RDF Schema terms

Table 3-7.

In addition to the basic techniques for describing resources using RDF statements, RDF users also need a way to describe vocabularies (terms) they intend to use in those statements. Specifically, vocabularies are needed for describing types of things, properties, and types of things that can serve as the subjects or objects of statements involving those properties. RDF provides no means for defining classes and properties. The basis for describing such vocabularies is the RDF Vocabulary Description Language: RDF Schema. RDF Schema offers a type system that was constructed with the type systems of object-oriented programming languages in mind. RDF Schema defines resources as instances of one or more classes which can be organized in hierarchies. For example, a class Convertible can be defined as a subclass of Sportscar which is a subclass of Vehicle. RDF class and property descriptions therefore give additional information about the RDF resources they describe. RDF Schema consists of a set of pre-defined RDF resources that have URIrefs starting with http://www.w3.org/2000/01/rdf-schema# which is normally bound to the prefix rdfs:. Consequently, vocabulary descriptions are valid RDF graphs. Table 3-7 gives an overview of the most important terms in RDF Schema. RDF Schema language elements

RDF Schema elements

description

example

rdfs:Class

A class corresponds to the generic concepts of type or category. RDF classes represent e.g., Web pages, people, documents or abstract concepts.

Vehicle, Car, Convertible

rdf:type

A standard property that defines that an RDF subject is an instance of a type defined in RDF schema. The property is also required for each class declaration to state that the class is of type rdfs:Class.

class “Convertible” is of type rdfs:Class, resource “BMW Z4” is of type “Convertible”

rdfs: subClassOf

This element specifies that a class is a specialization of another class. An application supporting the RDF Schema vocabulary can infer that an instance of subClass “Sportscar” is also an instance of “Vehicle”.

class “Sportscar” is a specialization of class “Vehicle”

rdf: Property

An RDF property defines a characteristic of a class using rdfs:domain and the range of values it can represent using rdfs:range.

power, numberOfSeats, registeredTo, roofType

rdfs: domain

This property defines which class(es) an RDF property applies to. An RDF property may have 0, 1 or more domain properties.

“power” is an RDF property of class “Vehicle”

3.3. Ontology-based Integration Table 3-7.

211

RDF Schema language elements

RDF Schema elements

description

example

rdfs:range

This property defines the allowed set of values and is used to indicate that the values of a particular RDF property are instances of a designated class.

all integer values; all persons defined in a class “Person”

rdfs:subPropertyOf

Properties can be specialized as well as classes. The RDF property in the subject is a specialization of the RDF property in the object.

“grossWeight” is a specialization of “weight”

Figure 3-11 shows an example RDF graph of a schema defined with RDF Schema, a class hierarchy for vehicles. There are five RDF triples in the graph with the subject being the specialized class, the rdfs:subClassOf element as property and the general class as object of the triple. Trucks, sports cars and passenger cars are specializations of vehicles. Convertibles are specializations of two classes, sports cars and passenger cars.

RDF graph

http://www.uibk.ac.at/iwi/vehicles#Convertible

http://www.w3.org/2000/01/rdf-schema#subClassOf

http://www.uibk.ac.at/iwi/vehicles#Sportscar

http://www.w3.org/2000/01/rdf-schema#subClassOf

http://www.uibk.ac.at/iwi/vehicles#Passengercar

http://www.w3.org/2000/01/rdf-schema#subClassOf http://www.w3.org/2000/01/rdf-schema#subClassOf http://www.uibk.ac.at/iwi/vehicles#Truck http://www.w3.org/2000/01/rdf-schema#subClassOf http://www.uibk.ac.at/iwi/vehicles#Vehicle

Figure 3-11. RDF Schema example “Vehicles” as graph

Listing 3-20 shows the class hierarchy specified in Figure 3-11 in RDF/XML notation using the typed node abbreviation. The URIs for the rdf and rdfs language definitions are declared and the base URI http://www.uibk.ac.at/iwi/schemas/vehicles is explicitly stated in the xml:base attribute at the beginning of the RDF/XML document. The individual classes are defined using the rdfs:Class element. Specializations are assigned with the rdfs:subClassOf property. The class Convertible has two rdfs:subClassOf properties indicating that the class is a specialization of both, sportscar and passengercar.

Classes and specializations

212

3. Integration Services

Properties

Three properties are defined in Listing 3-20. The rdfs:domain property defines which classes the properties are assigned to. The rdfs:range property is used to restrict allowed values. There can be zero, one or more rdfs:range properties to one definition of rdf:Property. For example, the property registeredTo is assigned to the class Vehicle and RDF statements using this property have instances of class Person as objects. In case of the power property, the rdfs:range property is used to indicate that the value of this property is a typed literal, here integer.

Listing 3-20. RDF Schema example “Vehicles” in RDF/XML

3.3. Ontology-based Integration

Datatypes are defined externally to RDF and to RDF Schema, and referenced by their URIrefs. The RDF Schema class rdfs:Datatype is used to explicitly state that xsd:integer is the URIref of a datatype. In the case of the property roofType, there is no range property meaning that the values of the property are not restricted. Listing 3-21 gives an example of an instance document that references the RDF Schema example “Vehicles”. The qualified name vehicles:Convertible becomes the full URIref

213 Datatypes

RDF instance referencing schema

http://www.uibk.ac.at/iwi/schemas/vehicles#Convertible and thus references the Convertible class in Listing 3-20. The two properties registeredTo and power reference the corresponding RDF properties defined in the schema. The schema describes the range of the power property as xsd:integer. Thus, the value of the property in the instance

document should be a typed literal of that datatype. The range declaration does not automatically “assign” a datatype to a plain literal. Therefore, a typed literal of the appropriate datatype must be explicitly provided. 140 Listing 3-21. Example instance of RDF Schema “Vehicles”

Summing up, RDF Schema extends RDF by a vocabulary for classes and properties similar to object-oriented programming. RDF Schema proposes well-defined rules for writing these definitions so that it is possible to access and exchange descriptions of complex properties and relations of Web resources. Descriptions can be parsed automatically to extract semantic information about Web resources.

3.3.3 Web Ontology Language (OWL) The Web Ontology Language (OWL) is used for defining and instantiating ontologies that include descriptions of classes, properties and their instances on the Web. The OWL formal semantics specifies how to derive the ontologies’ entailments, i.e. facts not literally present in the ontology.

Purpose

214

Revision of DAML+OIL

OWL compared to RDF (Schema)

Ontology as used in OWL

OWL’s three sublanguages

OWL Lite

OWL DL

OWL Full

3. Integration Services

Entailments can be based on multiple distributed documents containing ontologies that are combined using defined OWL mechanisms. One advantage of OWL ontologies is the availability of reasoning tools. Tools provide generic support that is not specific to any particular domain, thus easing the task of applying knowledge representation in EKIs to building domain ontologies rather than building reasoning systems. OWL is a revision of the DARPA (Defense Advanced Research Program) Agent Markup Language-Ontology Inference Layer (DAML+OIL). DAML and OIL are two XML-based languages stemming from a US DARPA program and a European Union program that have recently been integrated. OWL incorporates lessons learned from the design and application of DAML+OIL and is meant to be its successor. The normative OWL exchange syntax is RDF/XML, i.e. every OWL document is an RDF document. OWL has been designed for compatibility with RDF and RDF Schema. Compared to RDF Schema, OWL is a vocabulary extension that offers more facilities for expressing meaning, i.e. it adds vocabulary for describing classes and properties, e.g., relations between classes, cardinality, equality, richer typing of properties and characteristics of properties. OWL thus provides richer means to represent machine-interpretable content on the Web. Ontology is understood in OWL as formal definition and structuring of a set of terms and relationships to describe the terminology used in Web documents, thus representing the semantics of Web resources. Whereas the term ontology traditionally denotes definitions of classes, properties and relationships defining a certain domain, it is used here in a broadened sense and also includes instances and relationships between instances. Due to the fact that ontologies can be potentially useful in a variety of contexts, OWL provides three increasingly expressive sublanguages that cater to specific communities of users: OWL Lite, OWL DL and OWL Full. Each of these sublanguages extends its simpler predecessor in what can be legally expressed and in what can be validly concluded. OWL Lite contains a subset of commonly used features of OWL and limits their use. It provides simple constraints for the definition of classification hierarchies. The main reason for OWL Lite is that due to its lower formal complexity it is simpler to provide tool support for OWL Lite than for OWL DL and especially OWL Full. Also, OWL Lite supports migration for thesauri and other taxonomies. OWL DL supports all OWL language constructs, but limits their use so that computational completeness (conclusions are guaranteed to be computable) and decidability (computations are guaranteed to finish in finite time) are retained. DL is short for “description logics” and thus refers to OWL’s formal foundation. OWL Full removes many limits that OWL DL imposes and thus retains the syntactic freedom of RDF, however, with no computational guarantees. OWL Full is an extension of RDF, while OWL Lite and OWL DL are

3.3. Ontology-based Integration

extensions of a restricted view of RDF. Consequently, every RDF document is an OWL Full document. The choice between OWL DL and OWL Full mainly depends on the extent to which RDF Schema’s meta-modeling facilities are required. OWL Full allows to place restrictions on the vocabulary itself, therefore to augment the meaning of the pre-defined RDF or OWL vocabulary. In OWL Full, reasoning support is less predictable because currently there are no complete OWL Full implementations. As mentioned above, OWL extends the RDF Schema vocabulary. Table 3-8 briefly explains the most important added language elements. OWL Lite does not support all language elements, e.g., oneOf, disjointWith, whereas OWL DL and OWL Full support all language elements in Table 3-8. Furthermore, OWL Full allows arbitrarily complex class descriptions, consisting of enumerated classes, property restrictions, and Boolean combinations. Also, OWL Full allows classes to be used as instances, so that they can be used as class descriptions in one context and as individuals in another. OWL is explicitly designed for distributing and sharing ontologies. The owl:imports tag as part of the ontology headers provides a mechanism to include the entire set of assertions of another ontology. In order to minimize the effort required to build ontologies, many (standardized) concept definitions might be re-used with the help of this mechanism. OWL provides a number of language elements that specifically target integration of concepts defined in different ontologies. Examples are on the level of classes and properties equivalentClass, equivalentProperty, disjointWith, intersectionOf, unionOf and complementOf and on the level of individuals sameAs, differentFrom, AllDifferent and distinctMembers. This is especially useful when several systems storing parts of contents held in an EKI have to be brought together, e.g., several document management systems. For each of them, an ontology might be developed to capture concept definitions in this specific environment. These concepts then might be mapped with the help of OWL. It is a challenge to merge a collection of ontologies that have been designed with different application contexts in mind. The W3C’s OWL standardization effort is seen as encouraging development of tools for this task and for maintaining consistency between ontologies. OWL’s ability to link data from multiple sources described each with an ontology that is mapped to another source’s ontology is a powerful feature that can be used in many applications. However, the capability to merge data from multiple sources, combined with the inferential power of OWL, has potential privacy implications. A number of organizations are addressing these issues with a variety of security and preference solutions (e.g., OASIS Security Assertion Markup Language SAML, W3C platform for privacy preferences P3P, see also section 2.3.2, 126ff).

215

Complex classes

Sharing ontologies

Ontology mapping

Security issues

216

3. Integration Services

Table 3-8.

OWL language elements

OWL elements

description

examples

owl:Class

Just like RDF classes, OWL classes define groups of individuals sharing properties. Classes can be organized in a specialization hierarchy using rdfs:subClassOf. The built-in class Thing is the class of all individuals and is a superclass of all OWL classes. The built-in class Nothing has no instances and is a subclass of all OWL classes.

Vehicle, Car, Convertible class “Convertible” is a specialization of class “Vehicle”

owl:Object Property owl:Datatyp eProperty

RDF properties are further detailed according to whether they define binary relations between instances of two classes (object property) or relations between instances of classes and RDF literals or XML Schema datatypes (datatype property). Property hierarchies may be created using rdfs:subPropertyOf.

object: owns (Person, Car), powers (Engine, Car) datatype: make (Car, xsd:string), year-ofbirth (Person, xsd:integer)

instances represent individual members. It depends on the intended usage of an ontology whether a certain concept is considered a class or an individual.

“Audi A6” is instance of “Car”; in a different context, “Car” with license plate “I-4711” is instance of “Audi A6”

owl:Transi- If pair (x,y) and pair (y,z) are instances of the trantive sitive property P, then pair (x,z) is also instance of Property P.

ancestor (Bob, John) and ancestor (John, Bill) imply ancestor (Bob, Bill)

owl:Symmet- If pair (x,y) is an instance of the symmetric propric erty P, then pair (y,x) is also an instance of P. Property

partner (Ford, VW) implies partner (VW, Ford)

owl:inverse If property P1 is stated to be the inverse of property Of P2, then pair P1 (x,y) implies pair P2 (y,x).

registeredTo (Car, Person) implies owns (Person, Car)

owl:FunctionalProperty owl:Inverse FunctionalProperty

If a property is a functional or unique property, then it has no more than one value for each individual. FunctionalProperty is shorthand for stating that the property's minimum cardinality is zero and its maximum cardinality is 1. If a property is inverse functional or unambiguous then the inverse of the property is functional.

Functional: madeBy (Car, Manufacturer)

owl:allValuesFrom owl:someValuesFrom

This restriction requires that for every instance, if any, of a class that is related by the property to a second individual, the second individual is a member of the class indicated by the allValuesFrom clause. owl:someValuesFrom is similar and means that at least one second individual is a member of the class indicated by this clause.

class “Manufacturer”, onProperty “produces”, allValuesFrom “Car”, class “Course”, onProperty “isTaughtBy”, someValuesFrom “Professor”

InverseFunctional: produces (Manufacturer, Car)

3.3. Ontology-based Integration Table 3-8.

217

OWL language elements

OWL elements

description

examples

owl:cardinality owl:minCardinality owl:maxCardinality

The cardinality clause specifies the exact number of elements in a relation. minCardinality and maxCardinality refer to the lower and upper bound respectively. In OWL Lite, only 0 and 1 may be specified, in OWL DL and OWL Full, all positive integer values are permitted.

class “Car”, onProperty “yearLaunched” cardinality(1), onProperty “driver” minCardinality(1)

owl:hasValue

Classes are specified with this clause based on the existence of particular property values. A member of the class is required to have at least one of its property values equal to the hasValue resource.

class “DieselEngine”, onProperty “requiredFuel”, hasValue “Diesel”

owl:equivalentClass owl:equivalentProperty

The equivalentClass restriction indicates that two classes have precisely the same instances. The equivalentProperty restriction states that two properties relate one individual to the same set of other individuals.

class “Automobile”, equivalentClass “Car”; property “lectures”, equivalentProperty “teaches”

owl:disjointWith

This restriction declares two classes to be different. Thus, an OWL reasoning processor detects a conflict if an individual is an instance of two classes declared to be distinct.

class “Motorcycle”, disjointWith class “Car”

owl:sameAs

Two individuals are stated to be the same. OWL does not have a unique name assumption and thus two different names might refer to the same individual. A typical use of sameAs is to declare equality between individuals defined in different documents, as part of unifying two ontologies.

Author “Mark Twain”, sameAs Person “Samuel L. Clemens”

owl:differentFrom owl:AllDifferent owl:distinctMembers

differentFrom declares two individuals to be different, i.e. the opposite of sameAs. A number of individuals, especially a set of individuals, may be stated to be mutually distinct in one AllDifferent statement with distinctMembers declaring all members to be pairwise disjoint.

CarColor “Oceanblue”, differentFrom CarColor “Midnightblue”

owl:intersectionOf owl:unionOf owl:complementOf

These elements are used to define arbitrary Boolean combinations of classes and restrictions. These constructors specify the members of a set as in a definition.

class “GermanCar”, intersectionOf class “Car”, onProperty “madeIn”, hasValue “Germany”; class “EuropeanCar” unionOf class “AustrianCar”, class “BritishCar”...

owl:oneOf

A class can be specified via a direct enumeration of its members. The members of the class are exactly the set of enumerated individuals.

class “DaysOfWeek”, oneOf Day ”Monday”, Day “Tuesday”, etc.

218

3. Integration Services

3.3.4 Higher Levels of the Semantic Web Stack

Rules layer

Logical layer

Proof layer

Trust layer

Future of the Semantic Web

The higher levels of the Semantic Web stack are still under construction. The vision of Tim Berners-Lee, the founding father of the WWW and Semantic Web, is to provide a framework fully capable of exchanging logical rules, proofs as well as digital signatures using encryption and thus turn a web of reason into a web of trust. The most important difference to ontology-based systems which have long been known as expert systems is that this framework is envisioned to be able to handle a lattice of ontologies distributed within and across organizations and thus provide a strong mechanism for semantic integration in an EKI. In the following, we will briefly describe the vision of the layers beyond the ontology layer. On the rules layer, a limited declarative language should be provided that standardizes the way to query RDF statements. A rule language allows inference rules to be given which allow a machine to infer new assertions from existing ones. A comprehensive logic framework is meant to provide a vocabulary to fully describe and exchange logic assertions over the Web. The logical layer should turn a limited declarative language into a Turing-complete logical language, with inference and functions. The vision here is to provide a framework that allows for standardized sophisticated logic processing as specialized logic frameworks do today. Berners-Lee sees this language as being a universal language to unify all data systems just as HTML was a language to unify all human documentation systems. On the proof layer, applications or agents can share logic proofs. One agent can send an assertion together with the inference path to that assertion starting from assumptions acceptable to the receiver. This requires a standardized language and a standard proof engine. Berners-Lee suggests to build this engine on the basis of the logic layer or even on the basis of the less expressive rules layer. He sees a practical need for the exchange of proofs that might not want to wait until a sophisticated logic framework is standardized. The proof language together with digital signatures signing proofs should turn a web of reason into a web of trust. Thus, the logic framework inferences not only on logical assertions, but also on digital signatures. Berners-Lee has been optimistic about the kind of solutions that will be built using Semantic Web standards. If an engine of the future combines a reasoning engine with a search engine, it may be able to get the best of both worlds, and actually be able to construct proofs in a certain number of cases. The vast amount of data and meta-data on the World Wide Web leads to an overwhelming amount of logical processing that would be required to answer arbitrary questions about an unrestricted number of document sources. However, many specific real-life problems might be solved using inference engines that are constrained to a specific set of data, meta-data and logical assertions.

3.4. Function and Process Integration

Whether and when this bold vision will be fully embraced in commercial applications on the Web still remains to be seen. The Semantic Web initiative was for a long time accepted mostly in the artificial intelligence and agent communities, but has made progress to extend the community of users as the standardization process continues. XML, XML Schema and XSLT have gained wide acceptance and as a consequence are supported by a large number of tools. Compared to that, tool supply for creating, managing and using semantic meta-data created with RDF, RDF Schema and OWL is still in its infancy.

219 Tool support and commercial use

3.4 Function and Process Integration In contrast to data, meta-data- and ontology-based integration, there is another fundamental pillar of integration: functional integration and process integration that builds upon functions, brings them into a meaningful order and defines preconditions for their start. Functions are the atomic unit for interaction between information systems.

3.4.1 Function Integration There are several synonyms in use that express views on functions that different programming paradigms impose. The term function has roots in mathematics and has a strong focus on the result that is returned. In the procedural paradigm, the term procedure is used that focuses more on processing and is separated from functions that have to return a value. However, many programming languages use functions that return void, an empty value without a real data type, instead of procedures. In object-orientation, the term method is used for both, functions and procedures. Objects encapsulate data and methods that operate on that data. The term method refers to freedom of choice that the object has, whether it performs a requested operation or not, e.g., because of illegal parameters. Therefore, people sometimes speak of “sending a message to an object” instead of “invoking a method”. The agent-oriented programming paradigm further strengthens freedom of choice (section 1.5.1, 47ff). Another term with a meaning similar to function that is popular today is the term service (section 1.6.2, 69ff). From an integration perspective, only those functions are of interest that are explicitly made publicly available and thus can be called by other programs. The collection of all accessible functions is referred to as Application Programming Interface (API). An API consists of specification, implementation and documentation. Specification lists all functions and their parameters. Implementation usually is provided in form of a dynamic

Function

Procedure

Method

Service Application Programming Interface, DLL

220

3. Integration Services

link library (DLL) that encapsulates program code and can directly be used by other programs. The term DLL is mainly used in Windows environments whereas in Unix environments the term shared library (.so, shared object) is common. Documentation comprises a description of the function’s purpose, parameter values and intended effects. Good documentations also contain information about possible error cases. Besides integration, APIs are also commonly used for generating layers of abstraction that can be used to build upon. The operating system is a good example for that. It exposes APIs for drawing common screen elements like windows and buttons, file handling and communication. Examples for APIs used to integrate standard software are SAP ERP BAPI in the ERP domain and Open Text Livelink API in the EKI domain. Remote Procedure Call. Remote function calls are required in an integration scenario, as enterprise systems usually run on a dedicated server each. The term remote procedure call (RPC) is used for any kind of function call to software on remote systems. The idea behind RPC is to make a remote procedure call look as much as possible like a local one for a programmer. The goal is to simplify programming by relieving application developers from the need to open a socket, establish a network connection, pack and unpack parameters into a stream every time they access a remote system. This is performed by so-called stubs and the communication protocol used is often UDP. The client (calling program) and the server (called function) do not communicate directly with each other but instead have a stub each in their own local address space that is simulating the remote object. Figure 3-12 illustrates the functioning of an RPC. machine 1

1

machine 2 client stub

server stub

5 server

client 2

4 operating system

operating system 3

Network Figure 3-12. Principle of a remote procedure call (Tanenbaum 2003, 528)

Functioning of RPC

In step 1, the client calls the client stub. This call is a local procedure call, with the parameters pushed onto the stack in the normal way. In step

3.4. Function and Process Integration

2, the client stub packs the parameters into a message and performs a system call to send the message. Packing the parameters is called marshaling. In step 3, the kernel sends the message from the client machine to the server machine. In step 4, the kernel passes the incoming packet to the server stub. Finally in step 5, the server stub calls the server procedure with the unmarshaled parameters. The reply traces the same path back in the other direction. There are a couple of problems that prevent RPCs from behaving exactly like local calls. Passing parameters by reference certainly does not work, as client and server operate in different address spaces and therefore cannot share a data structure. The same reason makes it impossible to pass pointers to complex data structures like lists or arrays as the client stub has problems in determining where the structure ends and therefore cannot transfer it to the server. The client stub may also have problems in determining the data type of the parameters for weakly typed functions with varying numbers and types of parameters. Popular technologies for RPC on certain systems are Java RMI (Remote Method Invocation) and Windows DCOM (section 2.3.3, 137ff). As an example, we will take a closer look at Java RMI. In order to develop a Java program that uses RMI, the programmer has to design a number of classes and one interface that defines the methods that can be called remotely. The server part usually consists of two classes. One class implements the service interface and therefore provides the functionality requested by the client. In addition to that it implements the Remote interface to be capable of receiving RMI calls. The other class loads instances of the implementation and lets them listen for network connections. On the client, the calling class is needed and the client stub that is generated automatically by the RMIC tool that ships with the Java SDK (Software Development Kit). The client uses JNDI (section 2.3.3, 137ff) to locate the remote object. Listing 3-22 shows an example of how a Java RPC is implemented.

221

Limitations of RPCs

Java RMI

// Server RemoteClassImpl myImpl = new RemoteClassImpl(); Naming.bind("remote_class", myImpl); // Client RemoteClass rc = (RemoteClass) Naming.lookup("rmi://remoteserver/remote_class"); Listing 3-22. RMI code example RemoteClassImpl is the class on the server that implements the functionality the client needs. An object of this class is instantiated and it is being bound to a unique name. The client uses the interface RemoteClass

Object lookup

222

Serialization

XML over HTTP

3. Integration Services

to get a reference to the implementation on the server (RemoteClassImpl). The lookup is made using JNDI, but instead of an LDAP directory service, a tool called rmiregistry is responsible for the answer. It manages the names and bindings and has a function similar to the ORB in CORBA (section 2.3.3, 137ff). It can use CORBA IIOP instead of the Java remote method protocol (JRMP) to connect client and server stub. The default port for RMI communication is 1099. Besides instantiation of the object and additional Remote Exceptions that can occur each time a method of the remote object is called, the handling is identical to that of local objects. RMI even overcomes some of the RPC limitations listed above. Complex data structures in the form of objects can be passed as parameters to remote methods if they are serializable (i.e. implementing the Serializable interface of the Java API). Serialization in a nutshell means making a deep copy (all attributes with basic data types and copies of all objects referenced in attributes) of the object and creating a byte stream out of it. There also has to be the reverse mechanism that reads a byte stream and subsequently fills all the attributes with values. Web services. However, RMI and similar standards still have a limitations: they are bound to a specific platform. CORBA is one approach to solve this problem and provide language- and platform-independent RPCs, but due to complexity and initial performance problems, CORBA has not been adopted as quickly and widely as expected. Nowadays, a different approach gets more attention. Web services use XML data that is sent over HTTP connections to call remote functions. Two XML standards have been defined for Web services: WSDL (Web service Description Language) and SOAP (Simple Object Access Protocol). WSDL is used to describe the function, their parameters and the result and therefore is similar to an interface in Java or C# (Listing 3-23). … … … … … … Listing 3-23. Structure of a WSDL document

WSDL

A WSDL file consists of five sections that define types used, messages, portTypes, binding and service. The types section contains the definitions of data types that the Web service uses as input or output parameters. The XML Schema definitions (xsd:string, xsd:float, …) are used for the

3.4. Function and Process Integration

223

basic types. Complex types are specified using the XML Schema language. The message section defines name, input and output of functions that are involved. It can be seen as an API specification. The portTypes define so-called operations that use the messages as input or output. The operations distinguish four types that are listed in Table 3-9. The binding section maps the abstract definitions of the operations to concrete protocols, mainly SOAP and HTTP. The service section finally provides the URL where the Web service can be found. Table 3-9.

Types of WSDL operations waiting for response

without response

sending

solicit-response

notification

receiving

request-response

one-way

SOAP is used to wrap function calls including parameters into an envelope so that they can be sent over HTTP connections. The SOAP message consists of a header and a body section. As the contents of both sections are specific to the specific Web service, only a few things are specified. The elements in the header can use the three attributes encoding, actor and mustUnderstand to specify the encoding standard, which endpoint is addressed with the message (the final recipient or a station on the way) and whether the specified elements contain information that have to be understood by the receiver or that are optional. The body can contain a fault section if an error occurs. It describes the cause of the fault in a machine-readable (faultcode) and a human-readable format (faultstring). The faultactor element holds information about which part caused the error. Web services are sometimes also called XML-RPC, but one has to be careful as this term also denotes a Java technology that uses proprietary XML envelopes instead of SOAP to send data over HTTP connections. Java XML-RPC is still used sometimes, because it is much simpler than SOAP, although SOAP is the standard and is needed for interoperability with other Web services. Prominent examples for companies that offer services on the Internet using WSDL and SOAP are Google and Amazon. Web services can either be directly invoked by specifying the URL referencing their implementation, or they can be looked up in a registry, similar to CORBA or RMI registries. The standard for Web service registries is UDDI (Universal Description, Discovery and Integration). UDDI is XML-based like SOAP and WSDL and is regarded as the third pillar of Web services, although its use is optional. Originally developed by IBM, Ariba and Microsoft, it was handed over to the OASIS standards organization after the release of version 3 of the specification in July 2002. The goals of UDDI are on the one hand dynamic lookup and binding of Web

SOAP

UDDI

224

White, yellow and green pages

UDDI entity types

WS-Addressing

WS-Security

WS-Policy

WS-Attachments

3. Integration Services

services during run-time and on the other hand supporting developers in finding information about services, so that they know how to design clients that use these services. Originally, UDDI was meant to be a Universal Business Registry (UBR) on the Web where everybody could list and query Web services for free. However, more and more private UDDI registries have been set up by organizations that list only Web services within their boundaries and the development of the UDDI specification has reflected this evolution of purpose. The information in a UDDI registry can be categorized in analogy to a phone directory. White pages list organizations with contact information and the services the organization provides. Yellow pages list Web services classified according to a taxonomy. Green pages finally hold detailed information how a Web service can be invoked together with a link to the WSDL file. There are four main entity types in UDDI that are the basis for the descriptions. A businessEntity holds information about the provider of a Web service including name, address and contact information. A businessService describes a group of Web services that usually differ only in technical aspects like the URL at which the service is provided or different protocol bindings. A bindingTemplate holds technical information about how a service can be used, essentially the URL and parameters. The tModel finally is a container for any kind of specification like a WSDL file, a classification or a protocol. In addition to these three basic Web service standards, there is a number of other standards under development or that have recently been adopted. These standards extend or modify the Web service specification to address lacking mechanisms or bad performance. WS-Addressing extends the addressing capabilities of Web services to consider not only the target address, but also reference properties for routing of SOAP envelopes. These properties can specify additional relevant information, like a client id or session information. Reference properties store similar information as cookies do for common Web applications. More security was one of the most often demanded features for Web services. WS-Security introduces a new block within the header of a SOAP envelope that can hold authentication information (user name and password or a certificate), signatures or encryption information. It also specifies how this block has to be processed and builds on XML standards like XML-Security and XML-Signature. The WS-Policy framework allows specifying requirements for Web services that a client or server must fulfill so that both can communicate. A single requirement within such a policy is called assertion. Examples for assertions are authentication methods, transport protocols, or availability (e.g., 24x7). The WS-Attachments proposal and the “SOAP with Attachments” addendum to the SOAP specification describe how SOAP envelopes can

3.4. Function and Process Integration

225

be transmitted together with related documents like images, e.g., a scanned document, or technical drawings, e.g., a CAD file. DIME (Direct Internet Message Encapsulation) and MIME (Multipurpose Internet Mail Extensions) are used for encapsulating SOAP envelope and other documents. Other standards building upon Web services are discussed in section 3.4.3. Other integration technologies. Although Web services recently became the most popular approach to function integration, there are some other mechanisms worth mentioning. First of all, remote procedure calls do not have to be synchronous. Requests can also be put in a queue (section , 108ff) and processed asynchronously by the server according to the “publish and subscribe” model. But there are also some other forms of function integration. (Alonso et al. 2004, 33f) TP monitors are the oldest and best-known form of middleware. They are also the most reliable, best tested, and most stable technology in the enterprise application integration arena. Put simply, TP monitors can be seen as RPC with transactional capabilities. Depending on whether they are implemented as 2-tier or 3-tier systems, TP monitors are classified into TP-lite and TP-heavy monitors. TP-lite monitors typically provide an RPC interface to database, whereas TP-heavy monitors provide more functionality and can be applied in several cases. RPC was designed and developed at a time when the predominant programming languages were imperative languages. When object-oriented languages took over, platforms were developed to support the invocation of remote objects, thereby leading to object brokers. These platforms were more advanced in their specification than most RPC systems, but they were not substantially different in terms of implementation. Many object brokers use RPC as the underlying mechanism to implement remote object calls. The most popular class of object brokers is based on CORBA. Object brokers were designed to specify and standardize the functionality of middleware platforms. It soon became apparent that much of this functionality had already been available from TP monitors. At the same time, TP monitors, initially developed for procedural languages, had to be extended to cope with object-oriented languages. The result of these two trends was a convergence between TP monitors and object brokers that resulted in hybrid systems called object monitors. Object monitors are TP monitors extended with object-oriented interfaces.

3.4.2 Process Integration Like ontology-based integration builds on data and meta-data-based integration, processes build on functions and connect them in a shared context to support business processes. In the following, we first define the terms process and workflow and study types of workflows suitable for EKIs before we discuss workflow management systems, a system class used to

TP monitors

Object broker

Object monitor

226

3. Integration Services

define and execute workflows. Then, some standards are presented that build on top of Web services and coordinate their execution. Software implementations of these standards fundamentally change the way in which process management works in organizations with a move to service-oriented architectures (section 1.6.2, 69ff). We will conclude with an outlook towards semantic Web services that promise to join ontologybased, function and process integration. Definition of business process

A business process is (1) a closed, (2) repeating set of (3) interconnected tasks performed by (4) agents in (5) a temporal or causal order required to (6) fulfill a business function the aim of which is (7) to create value.

Analysis of the definition

(1) A business process can be separated from other activities in organizations. It has defined start and end points, typically specified by start events and end events. (2) A business process is conceptualized and modelled on the type level which means that there are instances, also called cases, that are performed according to the rules specified by the business process type. (3) A business process consists of individual tasks that are related to each other and thus need to be coordinated. (4) Human beings and machines are the two primary types of agents that perform tasks in a business process. A typical business process comprises tasks that are automatized, i.e. performed by machines, semi-automatized, i.e. performed by humans supported by machines as well as non-automatized, i.e. performed by humans without any machine aid. (5) The order in which individual tasks are performed within a business process is specified with the help of logical, time-based, geographical or quantitative constraints. (6) A business process is performed in order to fulfill a certain business function, e.g., design or production of a car, scheduling of flights which distinguishes it from other types of processes, e.g., the processes performed by an operating system as visible in the task manager of the Windows operating system. (7) Business processes also entail a certain perspective on organizations, i.e. the customer perspective. Ultimate goal of performing a business process is to create value for customers. This can be external customers that create business transactions exchanging goods and services for money or customers that are organization-internal, e.g., a core business process consumes services from a support process or a subsequent process takes over semi-finished goods from its predecessor. Business process and workflow management are important for organizations but focus that part of an enterprise infrastructure not considered in this book (section 1.5, 45ff). Knowledge work is characterized by unpredictable task sequences and complex tasks that cannot be described in sufficient detail with conventional process languages. Nevertheless, there is still a need for process integration in EKIs. Some parts within complex knowledge-intensive tasks can be automated or at least supported by

Process support for knowledge work

3.4. Function and Process Integration

227

workflows, e.g., publication of a document to readers interested in the domain which involves a knowledge broker responsible for checking the quality of the document and proper assignment to the company’s ontology. Table 3-10 characterizes three types of tasks. Table 3-10.

Types of processes in office work problem

information need

cooperation partner

solution

individual process

high complexity, low predictability

undefined

changing, not defined

open

case-based process

medium complexity, problem-depenchanging, medium predictability dent, partly defined defined

partly regulated

routine process

low complexity, high predictability

defined

defined

invariant, defined

Workflow Management. Routine tasks are best suited for workflow support but are seldom in knowledge work. Case-based tasks occur more frequently in knowledge work and can be supported by ad-hoc workflows. Individual cases play an important role in knowledge work and cannot be supported by workflows, but need different concepts (section 6.1, 408ff). Workflows automate business processes, in whole or part. They pass documents, information or tasks for action from one (human or machine) participant to another, according to a set of procedural rules. There are several views on workflows that stress different aspects: structured sequence of tasks: workflows define events, conditions and actions which determine the sequence of tasks, migrating objects: workflows can be understood as objects that are handed on from work place to work place, e.g., knowledge worker, supervisor, HR department for booking a formal training unit, updated conversational types: workflows can be seen as conversation nets where actors utter speech acts and other actors respond, e.g., an employee wishes to order a new computer and the department head rejects the purchase request, stateful objects: activities for a case can be seen as subsequent changes to the status of an object, e.g., a document that can be created, amended, reviewed, corrected, published and archived, rules: activities can be controlled by a set of rules to which they need to comply, e.g., “if volume>5,000 then authorization by underwriter”. Alternatively, workflows can be classified according to their structure:

Definition of workflow

Views on workflow

Level of formality

228

Ad-hoc workflows

3. Integration Services

Highly structured workflows are called standard workflows. Exceptions to these standard workflows that occur seldom but more than once can also be subsumed in this category. Semi-structured workflows are workflows that partly consist of teambased tasks or a number of tasks that are loosely coupled and do not always occur in the same sequence. Collaborative workflows are either completely team-based or involve ad-hoc planning of interaction between individuals. Furthermore, workflows can be used to roughly structure a complex process into several weakly structured and thus still complex tasks so that employees performing the tasks can check whether all relevant parts of the process have been fulfilled, e.g., systematic evaluation of a product. Onetime workflows are called ad-hoc workflows. In some cases, it is also reasonable to provide organization-wide templates with coarse granularity, e.g., for the proposal creation process of complex goods or services. Although most single tasks within this process would need further explanation and require a high amount of skills and experiences from the employee, it is still valuable to provide a framework that ensures that all parts of the process are completed in the right order and support the flow of electronic documents between the actors. Both, ordinary and ad-hoc workflows can be managed with the help of workflow management systems (WfMS).

Definition of WfMS

A workflow management system defines, creates and manages the execution of workflows through the use of software, running on one or more workflow engines, which is able to interpret the process definition, interact with workflow participants and, where required, invoke the use of IT tools and applications (WfMC 2002).

Build-time vs. run-time

An important distinction that has to be made for workflow management systems is between the two modes build-time and run-time. Workflow models are designed during build time. At run-time, instances of the workflow are executed, e.g., a concrete document instance is submitted to be published in a document management system and a reviewer is notified of the new document to check its quality and approve its release. Figure 3-13 shows the workflow life-cycle that starts with analysis and design of business processes which results in often semi-formal business process models. These models or parts of them are refined with sufficient details for workflow management systems. This happens during buildtime. Afterwards, the implemented workflow model is executed by the workflow engine. Execution can be monitored and logged in an audit trail. Periodically, audit trails should be analyzed to further improve workflows. This can also be seen as a learning process which results in another iteration of business process analysis and therefore re-enters the life-cycle.

Workflow lifecycle

3.4. Function and Process Integration

229

Workflows consist of actions (synonyms are tasks and functions) and events. Actors, which can be users of the WfMS or software applications for fully automated parts, are required to carry out the actions. Events can be seen as preconditions that need to be fulfilled so that an action can start or as results of actions. Boolean operators (and, or, xor), so-called connectors can be used to connect events with actions. Furthermore, objects of the workflow, e.g., forms or documents, are modeled as well as parameters required to invoke applications needed to perform individual steps of workflows. With these modeling elements, complex processes can be electronically reproduced, with actions that have multiple preconditions or different possible results. During run-time, the system watches for events that trigger a workflow. If all start conditions for the first action in the workflow are fulfilled, the system triggers the execution of the first action by either notifying the actor (often via email with a link to the workflow action in the WfMS) or directly invoking the function in the software application that can execute the action. Then, the actor carries out the action and reports the result back to the WfMS. The results then trigger one of the next actions and so on, until the workflow is finished.

Actions, events, actors

process knowledge

workflow controlling

business process analysis and design

process knowledge

business process model

workflow model implementation

workflow monitoring

audit trail data

workflow execution

implemented workflow

Figure 3-13. Workflow life-cycle (based on Mühlen/Hansmann 2003, 390)

Using WfMS has several potential benefits that are more or less applicable in concrete organizational settings: reducing waiting time: average time between completion of one workflow action and beginning of the next action can be reduced due to the notification the system sends, monitoring process status: enables real-time supervision of the states of all currently running processes, prioritizing processes: if a workflow needs to be completed as soon as possible, it can be flagged in the WfMS and therefore gets precedence over other processes that occupy the same resources,

Benefits

Design of workflows

Workflow engines

3. Integration Services

distributing work efficiently: task assignments are not bound to concrete persons, but to roles or groups, so that the workflow system can distribute work more efficiently using pre-defined rules, improving flexibility: WfMS encapsulate functions and data in a workflow for coordinating tasks that require multiple applications and thus are more flexible regarding changing processes and software upgrades. Design of workflows at build-time is often supported graphically. The Workflow Management Coalition (WfMC) is a consortium of companies that have interest in development and standardization of workflow products and techniques. XPDL (XML Process Definition Language) is an XML-based standard developed by the WfMC that can be used to exchange workflow definitions across applications and platforms. WfMC also developed an ideal architecture for WfMS (Figure 3-14). The core component is a workflow engine. In order to make the engine independent from supporting tools, a workflow API (application programming interface) has been developed and standardized. Therefore, workflow clients, applications that are invoked by the workflow engine, monitoring and definition applications have to comply to this API and can then be used with any workflow engine that implements the standards interface. process definition services

workflow API (WAPI) workflow enactment services workflow engine(s)

WAPI

administration & monitoring services

WAPI

230

other workflow enactment services

workflow API (WAPI)

workflow client application

invoked applications

Figure 3-14. Architecture of workflow management systems (based on WfMC 1995, 20)

Most workflow engines are embedded in other applications, e.g., Intershop Enfinity or IBM Websphere. Examples for stand-alone workflow engines are enhydra shark, Fujitsu Interstage Business Process Manager (iFlow), jBPM and WfMOpen.

3.4. Function and Process Integration

231

3.4.3 Web Service Composition A recent approach to increase flexibility of WfMS is based on Web services. The advantage of Web services compared to proprietary WfMS is that functions provided by the respective applications are described and invoked in a standardized way. Therefore, the limitation to a small set of supported applications that exists in all WfMS diminishes since a large number of developers of application system is expected to use Web services to describe the services offered by their applications. Coordination. A first step towards process management based on Web services (WS, section 3.4.1) is to coordinate their execution. The three standards WS-Coordination, WS-Transaction and WS choreography interface (WSCI) allow coordination and are briefly described in the following: WS-Coordination is a framework for supporting coordination protocols. It can be seen as a meta-specification that governs concrete coordination protocols. The framework specifies an extension of the SOAP header called coordination context for exchanging unique identifiers between coordinated Web services, as well as two interfaces. The registration interface is used to exchange information about the WSDL port of the WS. The activation interface supports exchanging information about the role of the Web service. One central or several distributed coordinators are responsible for coordinating Web services that participate in one workflow. During the activation of a Web service a coordination context is created and the following registration phase binds participating Web services to each other. However, the WS-Coordination standard does not describe how the actual coordination is implemented. This is the goal of WS-Transaction. Transactions that involve Web services are often long-running transactions in contrast to database transactions. Therefore, it is no option to lock involved resources over the entire transaction time. In addition to that, it is not clear what basic actions are involved in a transaction and how to roll them back in case of an error, e.g., how sending an email message can be rolled back. The result is that the ACID principle of data management which states that transactions should be atomic, consistent, isolated and durable, must be relaxed in the WS context. Consequently, a compensation operation is needed instead of a rollback. It is left to the implementation of a WS how compensation is performed. WS-Transaction specifies two protocols: business activities for long-running transactions as discussed above and atomic transactions for short transactions conforming to the ACID principle. Similar to start transaction, commit and rollback messages defined for database transactions, the business activities protocol specifies a start business agreement message that initiates a transaction, exited, completed and faulted messages that Web services can use to report states of their operations to the coordinator and close, complete, compensate and for-

WS-Coordination

WS-Transaction

232

3. Integration Services get messages which the coordinator can use to update Web services about

WS choreography interface

Composition challenges

Basic and structured activities

Orchestration

the status of the whole transaction. The Web service choreography interface (WSCI) basically defines four extensions to WSDL: exception handling: specifies what should be done if an exceptional situation is encountered, e.g., a corrupt message is received, transactions: only type of transaction (atomic or long-running) and corresponding rollback or compensate actions are specified. WS-Transactions is used for other aspects of transactions. correlators: in contrast to the unique identifiers in the SOAP header that WS-Coordination specifies, correlators are used to point to an attribute in the message body that holds the id. time constraints: different types of time constraints can be specified, e.g., time frame required between two consecutive invocations of a WS. Business process execution language. One level above these generic mechanisms to coordinate Web services are specifications describing Web service compositions, e.g., (1) request document meta-data, (2) order to purchase and download document, (3) confirmation and (4) delivery. Composition approaches need answers for the following questions: What types of basic components are composed? How are they orchestrated? How is data specified and accessed? How are concrete services selected? How are transactions handled? How are exceptions handled? There are several different proposals for Web service composition languages. The following paragraphs discuss how the specification with the biggest support by large IT companies answers these questions. Business process execution language for Web services (BPEL4WS or shorter: BPEL) is a specification proposed by IBM, BEA and Microsoft. It is based on XML and borrows many elements from earlier specifications such as the Web service flow language (WS-Flow) and XLANG. Basic BPEL components are WS defined in WSDL. These are called basic activities. BPEL processes can be reused as sub-processes in other BPEL processes as so-called structured activities. BPEL distinguishes synchronous function calls represented by the invoke activity and asynchronous function calls represented by the receive and reply activities. Furthermore, wait and assign activities are intended for blocking execution for a certain time frame and assigning values to variables. BPEL provides five constructs to orchestrate components: sequence executes activities in the defined sequential order.

3.4. Function and Process Integration

233

switch evaluates conditions associated with activities and executes the

first activity whose condition is true. It is also possible to specify an activity that should be executed if no condition is true. pick associates a set of events, such as receipt of a message with activities and executes the respective activity if the event occurs. while loop executes exactly one activity as long as the specified condition is true. flow starts a group of activities in parallel and is considered completed if the last activity is completed. In BPEL, it is possible to define variables similar to variables in programming languages. Variables have a name and a type and can be used as input and output parameters for invocations or as part of conditions. Variables can be of basic, simple or complex types as defined in XML Schema. XPath can be used to reference basic values within a variable of complex type. The assign activity can be used to initialize a variable or set it to a defined value at some state of the process. Selection of services in BPEL uses two constructs to determine a concrete WS that should be invoked. The partner link type determines roles involved in an invocation, relationships between the roles and the WSDL port types the roles have to provide. Partner links identify a so-called endpoint reference which is an abstraction of a concrete Web service. Endpoints can either be statically bound to a Web service by specifying a URL, or they can be dynamically bound by using a UDDI registry. Transactions are addressed by BPEL only in a limited way. An activity can be specified for compensation handling to undo effects of a successfully executed activity. Compensation handlers can also be specified for a scope. The default compensation handler for a scope simply calls compensation handlers of involved activities in reverse order of execution. It is expected that future versions of BPEL will build on WS-Transactions to provide more powerful mechanisms for transaction handling. BPEL relies on the proven “try, catch, throw” approach for exception handling which is known from several programming languages. Fault handlers can be defined either for one activity or for several activities by a scope. Faults are either generated by a WS using a WSDL fault message or by the execution engine, if a run-time error occurs. The throw activity allows to actively generate fault messages from within the process schedule. If a fault occurs, all currently running activities within the scope are stopped and the activity specified in the catch element of the fault handler is executed. In addition to the fault handler, an event handler can be specified that continuously monitors execution of activities and fires, if a certain event occurs. For asynchronous message calls and long-running transactions, it is important that messages can be connected to instances of processes, since it is possible that a process is instantiated several times at once. BPEL

Data types and access

Service selection

Transactions

Exceptions

Instance routing

234

3. Integration Services

allows to specify so-called correlation sets that point to message contents used to uniquely identify correlated messages. For example, an ISBN number could be used to correlate a book request with its response. There is still a lot of development going on in this area. The terms are not yet clarified, e.g., the terms choreography, composition and orchestration are often used as synonyms, and it is not yet clear whether BPEL will be the future standard for Web service composition, although it currently seems to be the most promising candidate. Similar standards like BPML (business process management language) specified by the BPMI consortium and XPDL from the WfMC are either less comprehensive or not equally well suited for WS. Products that already support the BPEL specification are BEA WebLogic, IBM WebSphere, MS BizTalk Sever 2006, Oracle BPEL Process Manager and SAP NetWeaver.

3.4.4 Semantic Web Services

Maturity of Web service composition

Maturity of semantic Web services

Integration developments

Semantic Web service is the keyword that denotes the vision to unify developments from the Semantic Web initiative and Web services, respectively composed WS in order to achieve more automation in the ad-hoc identification and orchestration of Web services. Web services are well standardized, several products that implement Web services are available and already in their second version. The early majority seems to be starting to apply these products in (parts of) their organizations. Web service composition, however, is still not as stable. There is no single standard, but several specifications from a couple of (important) software vendors, with BPEL being the most promising candidate for further standardization. First products have been available for a few years and innovators and early adopters are using it already in pilot projects or first operational settings. However, it seems like there is still a long way to go until broad adoption is reached. Semantic Web services are currently a vision. The need for an automated discovery and composition of Web services is identified. Although the term is used in an increasing number of publications (up from 25,000 hits in google in October 2004 to 500,000 hits in October 2008) and several working groups have been established, it seems to be still far from being ready for operational use. Especially the conjunction between OWLS (OWL for services), a promising technology for semantic Web services, and Web service composition languages like BPEL that do not get obsolete with OWL-S are unclear. Products are not yet available on the market. Figure 3-15 shows the development of data exchange and integration as well as of access to remote functions from the perspective of Web services and the Semantic Web. The goals of these approaches are ultimately that machines understand exchanged data and that services are universally interoperable. Steps towards these goals were taken under the headlines EAI (Enterprise Application Integration) and Semantic Web on the data

3.4. Function and Process Integration

comprehension axis as well as Web services and Web service composition on the service interoperability axis. Semantic Web services should integrate technologies from both development lines in the future. OWL-S uses OWL vocabulary to define a set of core constructs for describing properties and capabilities of WSs in unambiguous, computerinterpretable form. Four sets of data semantically describe WSs: service, service profile, service model, and service grounding. Service profile describes what the service does, service model specifies how the service does it (internal flow) and service grounding specifies how the service can be accessed (protocols). The service part then combines these three descriptions and must be referenced by the actual resource that offers the service (like the link to an RDF document describing an HTML page). Besides OWL-S which is an agent-oriented approach, providing fundamentally an ontology for describing Web service capabilities, there are other promising research efforts. An example is IRS-III (Internet reasoning service), a WSMO (Web Service Modelling Ontology)-based platform and infrastructure, backed by the European Semantic Systems Initiative (ESSI). These approaches need to consider multiple semantic dimensions that have to be described using ontologies: functional semantics to describe the function/effect of the Web service, data semantics to enhance inputs and outputs currently described in XSD, QoS semantics to describe non functional requirements a service has to fulfil, or criteria to chose between several suitable alternatives, e.g., time, cost, and reliability, execution semantics to describe preconditions to be met when calling the Web service, message patterns, flow of actions and results. service interoperability

semantic Web services

process integration

composed Web services

function integration

SOAP+WSDL+UDDI

isolated applications

proprietary formats

EAI

Semantic Web

XML + XSLT

RDF + OWL

isolated data

integrated data

integrated semantics

BPEL

BPEL + OWL-S?

Web services

data comprehension

Figure 3-15. Semantic integration and function integration (based on Daconta et al. 2003, 7)

235

OWL-S

IRS-III, WSMO

Semantic dimensions

236

3. Integration Services

Web service discovery and composition

These semantics can be used during Web Service discovery and composition. The following steps are proposed from deployment of the Web Service until it is executed by a service user: deploy => annotate => publish/advertise => discover => select => compose => invoke => monitor. When a service is deployed so that it can be used, it has to be semantically annotated in order to describe its use and make it detectable. Annotation also includes deciding which ontology to use and therefore is part of the ongoing ontology maintenance that happens in the background. Then, the semantic description of the service together with the technical description in WSDL can be published by a broker and thus the service is advertised for usage. If a client searches for a service it has to have a semantic description of what the service does (functional semantics) as well as of the desired input and output parameter (data semantics). With that information, a list of service candidates can be retrieved from the semantic registry using semantic matchmaking. Together with the service candidates, their QoS and execution semantics are returned which can be used to finally decide upon a service that matches best. Up to now, discovery of services often considers only one semantic dimension. Queries involving the service name, input, output and preconditions can be used alone for searching the semantic registry. Matching can also be done at the level of tasks or goals to be achieved, followed by selection of services which solves the task. This is done due to simplicity and well-known technology, e.g., from agent research, and in order to avoid overspecification that results in zero matches. Finally, the semantic descriptions could be used to support automatic composition and invocation of the selected Web service as well as monitoring its execution. All in all, there are a lot of open questions to be answered before the vision of automated discovery, selection, composition, invocation and monitoring of Web services can become reality. An additional obstacle for the development of semantic Web service standards is the lack of standards on higher levels of the Semantic Web stack. RDQL was approved only recently and there is still no standard for rules and trust. The connected uncertainty and fear to invest in a potentially non-viable technology is slowing down developments in the Semantic Web service area in spite of several interesting show cases developed in research projects.

Outlook

3.4. Function and Process Integration

Case Example Global Industries runs a DMS for the engineering department, a CMS as basis for the company’s Intranet and a collaboration platform for joint document creation and team work. Heidi has been assigned to integrate those systems to reduce redundancy and improve usability. In her role as project manager, she selects Ahmed, a SOA specialist, Shaun, the responsible employee for identity management, Clara, a specialist for XML and XSLT and hires an external consultant specialized in semantic technologies. In her project plan, the first step is to use the corporate LDAP-based user directory for single-sign on and user identification in all systems. The collaboration platform already builds on LDAP, but DMS and CMS have to be switched from their own proprietary user management to LDAP. Fortunately, the systems support LDAP as dominant standard, so that there is no need for custom developments, only configuration work. Shaun suggests to use Kerberos for authentication, because it is an open standard. Based on the central user directory, the company's telephone book and yellow pages should be replaced by a self-developed solution reusing information in the identity store and link phone records to user profiles in the collaboration platform that contain responsibilities, past projects, skills and pictures. The organization structure should also be visualized based on department and manager information already present in the user directory. Her second task is to standardize meta-data used to describe resources in all three systems. Her team comes up with a list of standard attributes that can be reused. These attributes like language, department, country, responsible person and validity period should have the same data type in every system. For usability reasons, the range of values must be customizable so that the list of choices only consists of relevant ones. Heidi's plans go even a step further. She wants to extend the functionality of all three IS to expose their contents as XML data that can be retrieved by Web services. The data can be displayed within other systems as HTML including meta-data and links to their original location. So Clara develops a number of XSLT transformations that can be reused in every system and display data as a list, show all document properties in a details view, or give a summary about document contents. Based on Web service technology, organization-external data sources can be integrated as well, e.g., to display documents in a map depending on their location meta-data. Finally, Heidi plans to introduce an inferencing engine to relate documents based on a central ontology. Ahmed suggests that the inferencing engine expose its services using SOAP conforming to the IT department’s SOA plans. The consultant warns her not to underestimate efforts to build a central ontology and suggests to develop a few division-specific ontologies and relate them with an integration ontology and OWL. Meta-data attributes can serve as a basis for concepts, classes and their attributes.

237

238

3. Integration Services

Questions and Exercises 1. What are the main goals why integration services are required as part of

EKIs? 2. What are meta-data? Give example meta-data for a digital resource of

your choice. Which meta-data standards could you use in order to support this task? 3. What are the main advantages and disadvantages of using standards for describing meta-data? 4. Reflect on the steps necessary to make use of meta-data in EKIs. 5. Give an overview of the most important elements and standards of the Semantic Web stack. What are the differences between XML Schema and RDF Schema? 6. Clarify the relationships between Unicode, DublinCore, XML and RDF with the help of concrete examples. 7. What is an ontology? Which types of ontology can be distinguished? 8. What is the difference between a taxonomy and an ontology? 9. Find examples for common sense knowledge to be applied in everyday life in Cyc and Open Mind Commons and compare them to each other. 10.XML files can be used as structured representation of initialization values or parameters for software applications. Get the file Vizx3D.ini from the supporting Web site26 and create an XML representation of it. 11.Develop an XML schema with local definitions for the following XML instance document: The teddy bear you love. Best seller! toys 9.99 14.99 stuffed animals Listing 3-24. XML instance document 12.Edit the result from the previous question by using global definitions.

Compare the two schemas and judge their suitability for this example.

26

URL: http://www.uibk.ac.at/iwi/EKI/

3.4. Function and Process Integration 13.Restructure the XSL transformation shown in Listing 3-5 so that it uses

only one template and two for-each loops. Then restructure it again so that it uses three templates and no for-each loop. 14.XSLT is often used to convert XML instance documents from one XML file format into another in order to exchange data between organizations. Get the XML and XSD files orderCyberware and orderXTerra from the supporting Web site12 and write an XSL transformation that converts the data from Cyberware format into the XTerra one. 15.XTerra sends delivery notes back to Cyberware to confirm that the ordered goods are on their way. Write an XSL transformation that converts the data from the XTerra into the Cyberware format. The source and target files can be found at the supporting Web site12. 16.Michael Meyer is a student and interested in knowledge management (KM). He has written a diploma thesis “Success factors for the introduction of KM”. This thesis is available at “http://www.michaelmeyer.de/uni/km-paper.pdf”. He participates in a community that deals with that topic and serves as a forum to discuss questions around KM. The community has a homepage at “http://www.brint.com/km”. The thesis was supervised by Prof. Dr. Reyam who has a homepage at http://www.reyam.de. Create an XTM topic map that represents the facts stated above. Pay attention to the correct choice of classes and instances, topics and occurrences, associations between topics as well as the roles that topics play in associations. 17.Make yourself familiar with the elements defined in the Dublin Core Metadata Initiative. Write an RDF file that describes the accompanying Web site of this book, using the Dublin Core standard. 18.Use the foaf-a-matic, an online tool available at the FOAF project (Friend Of A Friend) homepage http://www.foaf-project.org/ to generate an RDF representation of the description of yourself and your friends. Examine the results. 19.Make a list of the three internet applications you use most frequently (e.g., online shops, mail accounts, etc.) and write down the data each of them holds about the user. How much data is stored redundantly in each of the three applications? 20.Download the Web service toolkits from Google or Amazon and analyze the WSDL files. 21.Structure the information about users in profiles as part of account and identity management. What services of EKI can be supported by metadata on users of EKIs? 22.What are the most important goals that should be achieved by semantic Web services?

239

240

3. Integration Services

Further Readings A large number of books and articles deal with integration from a variety of perspectives. The standards touched in this chapter are welldescribed on the Web and can be easily retrieved from there. The following list contains works providing an overview of integration in general or specifically an overview of one of the distinctive integration areas discussed in this chapter. Berners-Lee, T., Hendler, J., Lassila, O. (2001): The Semantic Web. A new form of Web content that is meaningful to computers will unleash a revolution of new possibilities. In: Scientific American, May 17th, URL: http://www.sciam.com/article.cfm?id=the-semantic-web, last access: 2008-11-01 Tim Berners-Lee, the founding father of the WWW, presents his vision of the next WWW generation that gives an impression of the opportunities that a systematic treatment of semantics in the Web context could create. Daconta, M. C., Obrst, L. J., Smith, K. T. (2003): The Semantic Web. A Guide to the Future of XML, Web services and Knowledge Management, Indianapolis (IN), USA This book provides a comprehensive overview and discussion of the most important standards and technologies that are part of the Semantic Web initiative. Fensel, D. (2004): Ontologies: A Silver Bullet for Knowledge Management and Electronic Commerce, 2nd edition, Berlin Fensel provides an introduction into the concept of ontology and gives an overview of languages, tools and standards for developing ontologies as well as selected examples for their application in KM, business transactions and EAI. Fensel, D. et al. (2006,. 2008): Enabling Semantic Web Services; Implementing Semantic Web Services; Modeling Semantic Web Services, Springer, Berlin These are in fact three books that present the results of a number of research projects on Semantic Web Services resulting in an ontology, a modeling language and a technical framework to realize such services. Alonso, G., Casati, F., Kuno, H., Machiraju, V. (2004): Web services Concepts, Architectures and Applications, Heidelberg The book describes not only the basic Web service standards WSDL, SOAP and UDDI, but draws a complete picture including preceding technologies, middleware and EAI concepts as well as extensions to Web services for security, routing, coordination and composition.

4 Knowledge Services

Integration services as presented in chapter 3, 151ff represent the fundament that draws together contents from varying sources and offers a semantically and functionally, more or less cleansed and validated knowledge basis on which the core services can be built that differentiate an EKI from a basic information infrastructure. These services are classified according to the model of knowledge maturing explained in section 1.4.4, 42ff. Investigation services allow for searching and navigating the knowledge base which is not only restricted to explicated, documented knowledge, but also contains links to holders of knowledge, i.e. knowledge workers organized in organizational units such as teams or communities of interest. Individuation services up to now are the stepchild of knowledge services that can be found in EKIs. They allow knowledge workers to create personal contents, to appropriate contents created elsewhere and to manage their personal knowledge base, learning processes and career development. Interaction services target sharing of knowledge and its joint development in, e.g., teams, communities and knowledge networks. Information services target knowledge processes that need to move knowledge beyond the boundaries of context-sharing units such as communities and consequently require some formalization of knowledge. This includes, e.g., services helping knowledge workers to explicate knowledge including its context of development and potential application, document and content management as well as annotation using a controlled vocabulary. Finally, instruction services guide knowledge workers in enriching existing documented knowledge so that it can be taught and/or learned more efficiently and thus primarily help to prepare learning material so that it helps knowledge workers to internalize it, to examine whether these processes have been successful and finally to certify knowledge workers’ competencies.

Overview

On completion of this chapter, you should be able to give an overview of knowledge services and how they can support the different stages of the knowledge maturing process, give an overview of approaches for exploring structured data and explain which types of knowledge maps may support exploration of unstructured data, explain techniques for preparing contents for efficient search, how search results are ranked and visualized and how relevant information resources can be monitored,

Learning objectives

242

4. Knowledge Services

define the term individuation and describe services for handling, expressing and disseminating information about individual experiences as well as services helping knowledge workers to appropriate ideas, describe how individual competencies can be monitored, visualized and developed, explain the concept of expertise, stages distinguished in this relation and what follows for the design of EKIs, name and describe models to structure and formalize human communication, describe how EKIs can support creation of links between people, enable human communication and foster collaboration in communities, define the terms information resource, content, asset, document, document management and explain the role of document management systems, describe alternatives to capture and manipulate information resources, explain the advantages and different generations of Web content management systems, outline the needs for archiving information resources and name services used applied for this task, give an overview of learning paradigms and how they influence learning support by EKIs, define the term learning object and how these are composed to larger learning units, name ways suited to support consumption of electronic learning material and approaches to assess learning success.

4.1 Investigation Services Investigation services help to identify relevant knowledge in various forms and formats, particularly documented knowledge represented by various types of information resources. This is typically discussed in the field of Information Retrieval that deals with providing easy access to large amounts of information. It also relates to the fields of business intelligence and data mining as they research the exploration of structured data, enhanced visualization techniques for the browsing of knowledge resources and functions for monitoring knowledge sources. Figure 4-1 presents four groups of investigation services structured by means of a cycle. The first two types of services are based on a distinction between two general search modes. Exploration supports a process of retrieving information whose main objectives are not clearly defined in the

4.1. Investigation Services

beginning and whose purpose might change during the interaction with a system, i.e. if users lack knowledge about the domain or if they cannot express it precisely. An important way of support are knowledge visualizations, e.g., representations of sources of knowledge as well as their relationships. Search in contrast relates to a focused mode of search where a user needs to be able to describe the problem he deals with and ways to solve it at least to some degree, e.g., by formulating a search query consisting of a number of keywords. Presentation deals with the representation of search results, which particularly concerns their ranking, the visualization of relevant information as well as the obtainment of user feedback. Monitoring services support the continuous non-focused scanning of the environment and the gathering of useful “just in case” information. For example, selected electronic and paper-based resources are scanned periodically and potentially relevant information is gathered so that it can be easily retrieved when it is needed. exploration

monitoring

search

presentation

Figure 4-1.

Overview of investigation services

The general sequence of these services as shown in the figure indicates that an explorative mode of search can be seen as the first step for investigating information resources. The more knowledge is acquired about a topic, the more specific the information need and the required results can be formulated. Consequently, exploration is followed by a more focused search as well as by the presentation of potentially relevant results. Monitoring was positioned as the last step within the cycle as it is concerned with relatively specific topics whose development is observed over different resources. The dashed line between monitoring and exploration means that monitoring may trigger further investigation cycles when it yields new fields of knowledge that should be investigated. However, the sequence of services shown in the figure only should be taken as a general ordering. Principally, investigation services can be accessed in any order. In the following, we will discuss each of these services in larger detail. The following part introduces the field of information retrieval and related challenges (section 4.1.2) and afterwards a separate section deals with each group of services, i.e. exploration services (section 4.1.3),

243

244

4. Knowledge Services

search services (section 4.1.3), presentation services (section 4.1.4) and monitoring services (section 4.1.5).

4.1.1 Foundation Information retrieval

Understand user need

With increasing amounts of information, effective methods for information retrieval (IR) become more and more important. Searching large information bases has become a major issue in many organizations. Research on search algorithms, visualization techniques and text retrieval as well as vendors of software applications that packaged some of these methods into software products have catered to these needs. IR is the science of organizing and searching potentially large amounts of contents (Baeza-Yates & Ribeiro-Neto 1999, 1f). It has the goal to provide the user with easy access to information that fulfils his individual information needs. Meta-data is an important means for the identification of relevant documented knowledge as it is used for the classification and description of contents, e.g., by means of attributes such as author, title and keywords. IR traditionally focuses on documents and books that contain natural language. During the last years, research in IR has substantially broadened and also includes efficient indexing mechanisms, systems architecture, user interfaces, data visualization and filtering. Multimedia resources such as images and videos have became more and more relevant. Today, numerous tools help to support search, presentation and navigation of information and knowledge bases. Typical systems discussed in the context of IR are Online Public Access Catalogues (OPAC) adopted by many libraries (Rasmussen 1999, 407), e.g., of the common library network GBV or the Karlsruher Virtueller Katalog KVK that both integrate multiple other OPAC systems, full-text databases such as Lexis-Nexis or archives of newspapers and magazines such as the Electric Library (Rasmussen 1999, 399) and Web search engines such as Google and Yahoo (Glöggler 2003, 4ff). Challenges. As a foundation for this chapter, this section outlines a number of typical challenges that are investigated within the field of IR and need to be resolved by investigation services: First of all, investigation services have to understand the user needs and should hide implementation details. Both demands pose substantial challenges to designers of investigation services. Users usually must choose between search modes that determine how to find the information they need, i.e. exploration, searching and monitoring. A challenge for future research is to understand queries formulated in natural language and give concrete answers instead of presenting a number of sources that include some keywords. Currently, users often need to be familiar with the search strategy of the search engine in order to express needs as required and to achieve the results desired.

4.1. Investigation Services

Users do also want quick answers to their queries. Empirical results about Web searching behavior show that many users cancel queries that take longer than 3 seconds. Hence, some search strategies that would lead to better results cannot be implemented due to computational complexity and the resulting time required for delivering results. Certainly, the patience of a user depends on the actual search situation. Search processes in a business environment are partially different, e.g., because users might be willing to wait longer for business-relevant search results. Nevertheless, the quest for quicker search results calls for time- and resource-consuming preparation of the search domain, e.g., indexing and execution of searches on powerful machines together with caching of search results. Another challenge for search systems is to deal with a number of different sources. Relevant data could be stored on users’ PC, on servers in the Intranet or the Internet. It could even be stored on PCs of colleagues and ideally should be found there as well (see Groove example in section 4.3.5, 301ff). The same is true for documents in different formats or from different source systems. The more sources of information are included by the search system, the more comprehensive the results and consequently the higher the potential benefits will be. Next to text documents in varying formats, email messages can be considered as an important source of information, as well as human experts that should also be identified as part of a search for expertise within a domain. Furthermore, valuable information might also be stored in database systems or other enterprise systems that store data in proprietary formats (section 1.5.2, 56ff). Search engines should handle this variety of formats and sources without requiring active user decisions where to search. However, the more source systems are considered, the more complex and time-consuming is the query. As many advanced search strategies are based on linguistic models, the use of different languages within documents is another challenge. The languages of both user query and documents should be detected and ideally queries formulated in one language should also find documents in different languages. This is especially difficult as most words do not have a single but multiple possible translations and even in one language there are often several ways to express one idea. Context has to be considered, but it is missing in the query most of the time. That leads to recognition of semantic information, which helps to identify relationships between terms, e.g., synonyms, subclasses, homonyms or antonyms. For example, the search engine should be able to differentiate between documents that use identical terms in different contexts (homonyms) and - if not able to determine the right context based on the user query - at least group the results according to the different contexts it can identify. The structure of a document which is often given in a table of contents and headlines can be valuable to determine the context and identify interesting sections of the document, e.g., author or bibliography.

245 Quick search results

Different source formats

Language dependency

Semantic result processing

246

4. Knowledge Services

Number of results

Finally, users need the right information portion, e.g., the right level of detail or information corresponding to the user’s level of skills. Sometimes, a whole document is too much and a chapter is more valuable whereas in other situations the user needs not only a single email message or contribution to a newsgroup, but the whole thread or series of replies. An important problem for Internet search is the manipulation of documents to get higher ranked and therefore be more often accessed. This is referred to as search engine spamming. On the one hand, the search algorithm should be transparent to users so that they can use it efficiently, on the other hand the more people know about how ranking works, the better they can manipulate the results. Examples for manipulation techniques reach from applying meta tags to place keywords over texts displayed in background color or at a size of one pixel up to delivering manipulated Web pages to robots and letting other users access different pages. The latter can be done using one of the following techniques: Door pages are Web pages with redirects that hold content optimized for certain query terms. Cloaking is delivering different Web pages based on user agent identification. Hence, search engine bots such as Googlebot get delivered optimized pages different from that for actual end users accessing Web resources with browsers such as Mozilla or Internet Explorer. IP delivering is the technique of delivering different Web pages based on the IP address of the requestor. Similarly to cloaking, this can be used to present indexing servers with constant IP addresses other pages than normal users. As Internet search engines have developed into major starting points for Internet activities of many users, they have to carefully watch what information they spread. In addition to laws regulating spreading of information on the Internet, search engines also have to act socially responsible as they are powerful instruments affecting information exchange on the Internet. For example, providers of Internet search engines try not to index and present results from Web pages containing illegal or offending contents. Web servers hosting such pages are either manually put on a black list or can be identified based on a combination of keywords.

Search engine spamming

Social responsibility

4.1.2 Exploration EKI focus on the management of structured and semi-structured data. This distinction becomes very apparent in this section that starts out with the explanation of services focused on the exploration of structured data typically subsumed under the topic of business intelligence. Afterwards, it turns towards services supporting exploration of semi-structured data. Frequently, structured and semi-structured data both describe the same busi-

4.1. Investigation Services

247

ness objects from different perspectives. Hence, EKI need to strive for an integration of these classes of data. Exploration of structured data. Services that support the exploration of structured data are related to OLAP and to data mining. These two areas will be discussed in more detail in the following. The term On-Line Analytical Processing (OLAP) has been coined by Codd et al. (Codd/Codd/Salley 1993) in order to denote the provision of structured data for the support of business decisions. It stands in contrast with On-Line Transaction Processing (OLTP) which focuses on data processing in operational transaction systems. Whereas OLTP is mainly concerned with current and detailed data processed record-by-record according to the requirements of transactional systems, OLAP is based on aggregated and historical data and potentially involves a large number of records that are aggregated according to the specific analysis needs of business managers. It is included here as an investigation service as it assists the exploration of structured data by means of the visual presentation of multi-dimensional data cubes that are commonly projected to and displayed as two-dimensional tables. Edgar F. Codd, a pioneer of the relational model for databases, and two co-authors have formulated twelve evaluation rules for OLAP systems (Codd/Codd/Salley 1993). They state that an OLAP system needs 1. to provide multidimensional conceptual views on data, which actually is the central core of OLAP, 2. to be transparent, i.e. the user interface and the architecture of the system needs to be strictly separated so that details about the actual realization of the system are irrelevant from a user’s view, 3. to make external data sources accessible, i.e. the OLAP system acts as a mediator between the interface and various heterogeneous databases, 4. to be uniformly performant, i.e. it needs to have low response times also for increasing numbers of dimensions or larger database sizes, 5. to be based on a client-server architecture, i.e. in order to distribute the workload between system components one and ideally multiple clients need to be attached to a central server with minimum integration effort, 6. to support generic dimensionality, i.e. all dimensions have to be treated equivalent both in terms of their structure and operational capabilities, 7. to provide mechanisms for the efficient handling of sparse matrices, i.e. the OLAP system needs to automatically adapt its internal physical schema to the type of model, data volumes and sparsity, 8. to offer multi-user support, i.e. OLAP systems need to provide concurrent, secure and integrated access for multiple users that may retrieve data,

OLAP

Codd et al.’s twelve rules

248

4. Knowledge Services 9. to allow for unrestricted cross-dimensional operations, i.e. all forms of

FASMI

MOLAP, ROLAP, HOLAP

calculations should be possible across dimensions and not just selected ones, 10.to enable intuitive data manipulation, i.e. users need to be able to directly navigate based on the multi-dimensional data cube, 11.to offer flexible reporting functions, i.e. the dimensions need to be laid out in any way as preferred by the user, 12.to principally support ideally unrestricted dimensions and aggregation levels and if a maximum must be accepted, it should be at least 15 and preferably 20. This set of rules has been discussed controversial. For example, rule six (generic dimensionality) excludes that specific operational capabilities are granted to selected dimensions though this could be desirable, e.g., for the time dimension, or rule 12 (unrestricted dimensions) is technically not realizable and the numbers provided are arbitrary. Consequently, multiple suggestions have been made that change or extend this set of rules, which led to a large set of other proposals. One example is FASMI, an acronym for five generic rules that was coined in this context (Pendse/Creeth 1995). It stands for Fast Analysis of Shared Multidimensional Information: fast means that queries should not take more than five seconds on average, analysis requires that the system is able to cope with every logic needed and that more complex queries can be created with low programming efforts, shared refers to multi-user access and the realization of appropriate security mechanisms, multidimensional is the main criterion and stands for a multi-dimensional structuring of the relevant data and full support of dimension hierarchies, information means that the analysis must not be influenced by restrictions of the OLAP system which thus has to be independent of the amount and sources of the data. OLAP systems can be categorized based on the underlying data schema. Multidimensional OLAP (MOLAP) stores data in a multidimensional array storage which allows enhanced performance due to optimized storage, multidimensional indexing and caching mechanisms. Relational OLAP (ROLAP) is based on databases that hold the data in relational tables. Aggregated data is stored in additional relational tables for combinations of the relevant dimensions. In order to ensure an appropriate level of performance, specialized relational database schemes are applied for the modelling of the relational data, e.g., the star schema or the snow flake schema. One of the advantages is that it can be accessed by traditional

4.1. Investigation Services

249

SQL tools. The term hybrid OLAP (HOLAP) is used in order to denote databases that combine relational and specialized storage. After an initial success of MOLAP systems, established RDBMS caught up so that today ROLAP and HOLAP systems prevail. Table 4-1 presents a set of OLAP operations. They are clarified by means of an example for a company that has data about its turnover detailed based on the four dimensions, geographical region, product, customer and time, which represents a four-dimensional data cube. The operations essentially describe transformations on this data cube. Table 4-1.

OLAP operations

operation

description

example

slicing

select a layer from the data cube (two-dimensional projection)

Time and customer are selected in order to analyze the turnover. They are presented as a two-dimensional table with time on the x-axis.

dicing

change horizontal and vertical projection of tabular presentation of the data cube

The dimension customer is exchanged with region in order to analyze the turnover from this perspective. Hence, the vertical projection of the table is changed.

drill-down / roll-up

change to lower / higher levels of granularity

The dimension region is detailed into single districts which enables a more fine-grained analysis of the turnover (drill-down).

ranging

restrict range of dimensions

The turnover is displayed only for one product category which substantially reduces the number of facts to be analyzed.

split / merge

adds / remove dimensions to/ from the data cube

The dimension customer is added in order to analyze the turnover according to time, region and customers.

drill-through

drill-down operation associated with change of data source

Current data is included from operational transactional systems.

drill-across

report current key figures resp. data of another qualification hierarchy

The multi-dimensional table is complemented with key figures such as the return-on-investment in order to enable comparisons.

Data mining is the systematic process of discovering patterns in a defined data set by applying methods of explorative data analysis. Knowledge discovery in databases (KDD) goes a step further and points out that the result is new knowledge, which explains why it is important to con-

Data mining, knowledge discovery in databases

250

Knowledge discovery process

4. Knowledge Services

sider in EKIs. It is typically applied when the selection of an applicable statistical model cannot be definitely predetermined and when the initial question cannot be clearly defined. This is the main difference to traditional statistical analysis approaches. Some authors add that data mining analyses large amounts of data as this may be a precondition for the identification of statistically significant results and which is one of the reasons for the application of computer technology. However, this attribute should be used with caution since “large” is a very relative term that requires further detailing. A more important characteristic is that data mining services are integrated for the seamless support of (parts of) the knowledge discovery process, which is described in the following. This is another feature that can be used to delimit data mining from isolated methods used for statistical analysis. The knowledge discovery process can be described by means of the following five steps: 1. Data extraction, transformation and loading is conducted in order to integrate operational and potentially heterogeneous data sources. Heterogeneous means that they are stored within diverse systems and may have individual logical data schemes and formats that are optimized to serve the operational transactions that the specific applications are designed to serve. Therefore, the data needs to be extracted from these systems and transformed to integrate it into an appropriate data scheme, e.g., as part of a data warehouse. Transformation includes (1) filtering relevant data, (2) cleaning of data, i.e. removal of errors, (3) harmonization, i.e. assuring consistency of formats and meaning of data elements, (4) aggregation, i.e. summarizing data, e.g., for longer time periods and (5) enriching data, e.g., by business ratios. Loading is duplication of relevant data and its storage within a separate database. 2. Data selection and dismissal of extreme cases includes the definition of the tables and attributes that are relevant for analysis, e.g., by implementing a database view, as well as the removal of cases that may distort statistical results, e.g., by excluding extreme values. This step depends on the method to be applied for data mining and may also include the aggregation of data in order to enable the efficient application of mining algorithms. For example, log-entries that trace the phone connections to a call center for every second are aggregated to hours or days as a more fine-grained resolution of the data is not required nor manageable. 3. Hypothesis generation involves the scanning of the portion of the data to be analyzed and the creation of hypothesis that are relevant with regard to the goals of analysis. Naturally, a general preconception about the goals and sometimes also the relationships to be tested already existed in preceding steps as this also influenced the selection of the data to be analyzed. This step is strongly based on knowledge about the

4.1. Investigation Services

data and its structure as well as about its context represented by knowledge, e.g., about the specific industry or domain area to be investigated, and thus depends on the competencies of the individual analysts. 4. Application of data mining techniques is the execution of algorithms on the data in order to identify patterns, regularities and associations. The choice of the technique depends on the given tasks that should be solved during analysis. Typical tasks are classification, segmentation, association analysis, prediction and variance analysis. Techniques applied are neural networks, decision trees, clustering and association methods. For a specific task, one or multiple techniques may be appropriate depending on the concrete circumstances. Segmentation for example can be achieved by means of statistical clustering algorithms or neural networks. Table 4-2 give a condensed overview of the most important data mining tasks. Table 4-2.

Data mining tasks

task

description

example

classification

assignment of objects to predefined groups based on the similarity of selected attributes

customers are classified into four pre-defined risk groups AD in order to estimate the probability that they claim insurance payments

segmentation

summarization of objects into groups that are not known in advance

definition of customer groups based on their preference of specific product characteristics

association analysis

analysis of relationships between object characteristics of one and the same or of different objects

consideration of products that are typically bought in relation to other products, e.g., milk, butter and bread are frequently bought together

prediction

forecasting of future values based on a logical series of given values

estimation of future stock value of company shares on the stock market

variance analysis

identifies objects that do not follow regularities assumed or identified in relation to other data mining tasks

groups of people are compared in terms of the degree to which specific personal characteristics influence their amount of TV consumption

5. The presentation of results comprises the conversion of the results into

a model that can be easily interpreted by humans. Examples for models are diagrams, formal languages as well as natural language sentences. This step is important since the support of human perception and interpretation ultimately determines the success of the whole knowledge dis-

251

252

OLAP vs. data mining

Knowledge source maps

4. Knowledge Services

covery process. However, the interpretation of the results ultimately depends on the skills of the human user. This step may lead to the creation of new hypotheses or to the redefinition of existing ones which leads to a repeated iteration of the knowledge discovery process. Though data mining and OLAP have similar goals, they differ with regard to the data basis, the role of the human user and the type of user group. Data mining tools typically operate on a data basis that is already prepared for this task. This is due to specific requirements, e.g., for structuring of the data and for optimization in order not to cause performance bottlenecks. The two fundamental understandings of the user’s role can be contrasted by means “human-assisted computer discovery” vs. “computerassisted human discovery”. In the first case, humans support computer systems by providing answers to specific questions and thus help to compensate weaknesses of IT systems. The second case describes the circumstance that humans do the analysis mainly on their own and computers assist them, e.g., by processing large data amounts and offering different forms of visualization that may enhance human creativity. Though the boundaries are blurry, data mining services can be classified into the former class whereas data warehouses and OLAP services mainly belong to the second class. Furthermore, whereas data warehouses typically are prepared in order to support employee self-services for reporting by many employees, data mining typically is used by a much smaller and mathematically and statistically competent user group. Exploration of semi-structured data. Besides the exploration of structured data, an EKI also offer services that enable for an overview and investigation of semi-structured data, e.g., by offering means for browsing and navigating through potentially large document collections. In the following, concrete implementations of IT-supported knowledge maps will be presented for two important types of maps: knowledge source maps and knowledge structure maps. Afterwards, the Infotop prototype will be described as it offers innovative means for visualization and navigation within large collections of documented knowledge. Furthermore, recommendation services will be outlined as they offer means to discover new knowledge. Knowledge source maps are designed to assist the identification of the location of different forms of knowledge, e.g., people and their competencies or information systems that store documented knowledge, and their relation to knowledge domains or topics. The visualizations may also include further information such as the amount and complexity of knowledge. Figure 4-2 presents an example. The map visualizes what experts are available for the company’s five competence areas animation, database, graphic design, project management and technology know-how and the three product lines Web systems, stand-alone systems, CD-ROMs. Furthermore, it contains information about the assignment of these employees

4.1. Investigation Services

to the three main locations Basel, Berlin and New York or whether they are mobile and travel between the three locations. This map may be applied, e.g., for the staffing of multimedia projects that may involve employees from multiple different locations and their distinct competencies.

Diane Strong

Graphic Design

Animation

Anne Weick

Ina Roehl

= New York

Marion Pressl Eva Rohner

Steffi Sieger

Max Hitz

Karl Toner Patrick Auer

Stefan Werd Alex Müller

Technicians Ute Lemp

= Berlin

Thomas Schmid

Julia Venn

Database

Holger Stier Uli Rubner

Maya Senn

Web

Michael Gross

Martin Sik

Maria Galatea

= Basel

StandAloneSystems

CD-ROM

Chris Teiler

= mobile

Mark Ott

Project Josef Goner Management

Figure 4-2.

Example for a knowledge source map (Eppler 2003a, 195)

Figure 4-3 illustrates sources for project knowledge using the metaphor of a tube map. www.brint.com

www.cio.com

KMReferen ce Model

G. Nittbauer (Kon stanz)

KM Case Book

www.lu.un isi.ch/n ewmine

Project Plan VKBAnalysis

Lug an o NSF Prop osal

BBAn al ysis

Case Stud ies

NetAcad em y. org

D. Diemers

Final Report

NA Ikarus

Strat-DB Interactive BSC

MCM Zentral

K-Mapper

Qu ality Standards

EKM2-DB

S. Seufert Coral

M. Franz (Siemens) Peter Stadelmann

T ct

ne eli im

Exp ert Directo ry

IQRadar Marent. com

ASCO PM T rend s Commun ities

Knowledge Portal On Track 3

B. Huber (UNIZH)

Key Learning

Decision Discovery

Key Account Portal

Knowledg e Brand s T eam Knowledg e Managemen t KM-Suites Evalu ation

O. Christ

Competence Man ag emen t

AktivPlan

EKM1-DB Daedalus Domino Group Do c Project

Elias

eBaraza.ch www.cck.u nikl.d e/wmk

Knowledge Media.org

Informationquality.ch

www.Expertsexchange.com

Pro cess Knowledge

Figure 4-3.

Müller (ThinkTools)

Peak

Renaissan ce

oje Pr

Facilitatio n To ols

K-Maps

www.mcm.u nisg.ch DCAn alysis

Heiko Roehl

KM Glossary

Project Knowled ge

R. Will (Ring ier)

Token

Ron Hyams

Experts Documents Applications Websites Databases Publications

Outside Experts Drafts Legacy Apps. Dead Links Archives Grey Literature

Example for a tube map used for knowledge identification (Eppler 2008, 136)

253

254

Knowledge structure maps

Example: Selforganizing map

4. Knowledge Services

The map covers e.g., experts, documents, applications or Web sites. Dashed lines are used for knowledge source that has specific characteristics, such as being in a pre-mature state, i.e. drafts, potentially outdated knowledge, such as stored in archives, not as easily accessible knowledge such as found in external experts or in grey literature. The map gives a good example for applying a collaborative mapping process rather than an automatic mapping process. The communication between the knowledge workers involved in building the map are used for collaborative sensemaking, for identity- or consensus-building, here, in a project team. Thus, knowledge maps can also be used as prime tools for interaction (section 4.3). Knowledge structure maps show relationships between different knowledge domains or topics and should not only visualize that there is a relationship, but also explain the type of relationship such as generalization, aggregation, synonym or antonym. A good example for knowledge structure map is the self-organizing map by Kohonen (1995). The author is the inventor of the algorithm which automatically clusters input documents according to similarities between vectors of concepts they contain and forms the basis for visualization in the map. Each distinct topic or cluster of documents is visualized by a separate mountain or hill. Geographical proximity in the map represents close relation of the clusters or topics. Deep valleys between mountains visualize that the concepts or documents on the two mountains are considered semantically distinct. Figure 4-4 shows an example of a self-organizing map that represents news. The keywords are extracted using preparation techniques explained in section 4.1.3.

Figure 4-4.

Self-organizing map showing news1

4.1. Investigation Services

255

A common way for visualizing the structure of knowledge are hierarchical classifications such as hyperbolic trees. They use the root node in a hierarchy as the central focus and groups node that are directly related to this node in the focus around it. Each related node has other related nodes that are positioned further to the outside of the display. A kind of “fish-eye view” can be used in order to utilize space on the screen more efficiently. In this case, the few nodes in the center are displayed using a comparably large font and the many nodes at the outskirts of the graph are displayed using a smaller font. Figure 4-5 shows an example of a hyperbolic tree that is used as navigation structure for a Web site. It was generated with Inxight VizServer.

Example: Hyperbolic tree

Figure 4-5.

Hyperbolic tree representing the structure of a Web site2

The research prototype Infotop (Maier/Sametinger 2004) uses an innovative approach for visualization. It does not represent a specific type of knowledge map but rather offers a general approach for the navigation through large collections of information. It is based on the multi-dimensional classification and visualization of items as done in the OLAP 1 2

URL: http://mappa.mundi.net/maps/maps_015/, last access: 2008-11-01 URL: http://www.sapdesignguild.org/community/design/hierarchies4.asp, last access: 2008-11-01

Multi-dimensional presentation

256

4. Knowledge Services

domain. Therefore, it distinguishes six general dimensions that can be regarded as classes of meta-data: time, topic, location, person, process and type. Not all dimensions are equally important for all types of items, but most of them apply in any case. Each dimension may have sub-categories for a given item type, e.g., the dimension person can be further detailed into senders and receivers for email messages. Consequently, Infotop allows to apply OLAP operations such as slicing, dicing, drill-down and roll-up in order to dynamically navigate through collections of information items. In a simple case, selected dimensions can be used in order to filter relevant items. In Figure 4-6, an example is shown that uses person and process to form a two dimensional view on documents of the type text (see section 3.2.3). The pile metaphor is used in order to indicate the amount of results within a specific cell.

Figure 4-6. Recommendations

Multi-dimensional visualization with Infotop

Recommendation services try to suggest new items to the user that are of potential relevance or interest for him. MovieLens3 is an example for a research prototype in this area that generates suggestions for movies that a user may like, DVDProfiler4 is an example for a commercial product with comparable functionality. Amazon.com uses recommendations such as “customers who bought this book also bought these titles / titles by these authors” in order to suggest potentially interesting products to customers. Recommenders may be based on content-based filtering or collaborative filtering. Content-based filtering analyzes and selects contents based on content-related criteria such as pre-defined key words, similarity to other contents the user has accessed or more advanced methods like neural networks that represent evolving interest profiles of users. Content-based fil3 4

URL: http://www.movielens.org, last access: 2008-03-03 URL: http://www.invelos.com/Products.aspx, last access: 2008-11-07

4.1. Investigation Services

tering selects items because they are similar to items a user liked in the past. Collaborative filtering in contrast identifies and selects (groups of) users whose tastes are similar to those of the given user and then recommends to him what these users have liked. As a consequence, it is possible for a system to identify and recommend items that are dissimilar to those that a user has seen in the past or that he has specified by means of a search query. This approach is effective when the number of users is large compared to the amount of content in a system. If the number of users is small in relation to the volume of information, there is the danger of the coverage of ratings becoming sparse, thinning the collection of recommendable items.

4.1.3 Search In contrast to exploration, search is a directed method to retrieve specific information specified by query terms. Figure 4-7 gives an overview of steps that are supported by search and presentation services. Instead of searching through all resources in order to answer a search request, an index is accessed which substantially enhances the efficiency of the search process. The steps acquisition & pre-processing as well as indexing thus represent important preparation steps that are concerned with the creation of an index of the search resources. Requests are formulated in the form of search queries that need to be analyzed and understood by the search system before any results can be delivered. A user profile may assist the identification of relevant results. This section will explain these three steps in more detail. Finally, it will outline a number of different types of search services. The further steps included within the figure are targeted towards the presentation of search results and thus are described in the next section (section 4.1.4). search services

presentation services

1. acquisition & pre-processing

4. ranking index 5. result visualization

2. indexing 3. query analysis

user profile

specify/refine query

Figure 4-7.

Steps supported by search services

6. feedback

navigation of results

257

258

Function of a crawler

Conditional GET

Overview of pre-processing steps

Lexical analysis

4. Knowledge Services

Acquisition & pre-processing. The most common way to acquire resources for indexing is using a crawler. A crawler is a program that browses the Intranet or Internet in a systematic and automated manner. Synonyms for crawler are Web crawler, spider, Web robot and Web wanderer. Besides the different names, they all follow the same principles. A crawler consists of gatherer, loader and checker components as well as a URL database. Seven consecutive steps describe its function: 1. Starting point is a list of URLs, folders on file servers or other sources of information. 2. The loader assigns tasks to visit information sources to the gatherers based on the list. 3. The gatherer collects the data using http requests or other protocols and hands them over to the checker. 4. If the returned resource (e.g., document, html page) contains hyperlinks or the directory has subdirectories, the gatherer appends the new URLs to the URL database. 5. If the URL is not pointing to a resource (any more), it is deleted from the database, otherwise it is marked as visited. 6. The checker examines the resource and decides whether it should be indexed. In order to accomplish this, it first checks the resource’s format (e.g., html, pdf, doc, rtf) and looks for a corresponding filter that can extract plain text from the resource. Then, it checks if the resource has already been indexed (doublet-check). 7. If all checks are o.k., the content is handed over to the indexer. In the common case that HTTP is used as access protocol, a conditional GET request can be used to reduce network traffic. The conditional GET statement can contain parameters so that the requested resource is only returned by the Web server, if the condition is fulfilled. The most important condition is the modified date. As the indexing task has to run periodically (daily or weekly), the crawler can specify the date of the last indexing as parameter and the document is only transferred if it has been modified since then. Otherwise, the HTTP status code 304 (not modified) is sent back instead of 200 (ok). Once the contents of the search domain are determined, the resources must be prepared for indexing. In the following, we will briefly review this process for text documents which converts the document into a representation of (parts of) its contents. This includes lexical analysis, language detection, elimination of stop-words, stemming and the identification of noun groups. The first step is lexical analysis. The purpose is to convert the stream of characters in the document into a stream of words. What seems to be an easy task as space characters delimit each word is quite difficult at closer inspection. One challenge is how to deal with digits. On the one hand,

4.1. Investigation Services

numbers usually are not useful search terms and therefore are often neglected during indexing. On the other hand, numbers that represent years, prices and sales numbers can be important for document retrieval. Another decision has to be taken concerning how hyphens should be treated. If the composed word is decomposed into its elements a part of the semantics can be lost. Dots usually separate sentences and can safely be removed if there is no further analysis on sentence level, but again a few exceptions as DNS names or variables in object-oriented programming can have dots as integral part of their names. Finally, the lexer (short for lexical analyzer) converts characters to either lower or upper case as the search is usually case-insensitive. The language of the document has to be determined before elimination of stop words can be applied. That can be done either by using linguistic analysis (generic) or based on comparison of a few randomly chosen words with dictionaries (one for each supported language). Language analysis can be done on the level of documents. However, this is not sufficient in cases where documents contain paragraphs or even single words in different languages. Before the document is being added to the index, some improvements have to be taken in order to make the index more effective. First of all, a list of stop words is processed and all words included in the list are removed from document representations. The reason for that is that words that are present in nearly all documents are not suitable as index terms. Examples for stop words are is, a, and and to. In general, words should be eliminated that are not discriminatory, i.e. are not suitable to distinguish different documents. Candidates for stop words are articles, prepositions and conjunctions. A second advantage of this method is that index size can be reduced by 40% on average. Another technique that improves index quality is called stemming. It reduces words to their stem by removing pre- and suffixes, so that nouns are all singular and verbs are all in their base form. Most stemming algorithms can only deal with regularly built forms, which is sufficient for most words, but can lead to unwanted results in special cases. Again, a trade-off between fast processing and best results has to be made. Finally, in some cases it makes sense to further reduce the number of index terms. Almost all keywords used regularly are nouns. That is because nouns are usually more distinct than verbs. An algorithm for automatic index term selection can make use of this knowledge and searches the text for nouns and especially for nouns that are grouped together within a sentence. These noun-groups are identified based on their syntactic distance in the text, measured by the number of words or characters between them. So, noun-groups are used as index terms instead of single words.

259

Language detection

Elimination of stop words

Stemming

Noun-groups

260

Inverted list

Text compression

Index merging

Query types

4. Knowledge Services

Indexing. An index is a data structure that contains redundant data describing and/or pointing to text. It is used to speed up search, as otherwise online search of the whole document set has to be done for every query which is inefficient and not feasibly for texts larger than a few MB. Indexing is only a question of the most efficient form of storing index terms and their occurrences, as a lot of work has already been done in preprocessing. The most commonly used index structure is an inverted list (also called inverted files). Index terms are sorted alphabetically (the vocabulary) and point to where these terms appear in the documents (the occurrences). The occurrences can be expressed in terms of characters, words or blocks of a certain size (e.g., 16 or 64 kB). The finer the occurrence is saved, the more memory it needs. Therefore, it is often accepted that a small portion of the text has to be inspected by users in order to determine the exact position. Indexes using blocks of 64 kB take up only 5% to 9% additional storage space for large text collections (e.g., 2 GB ASCII text) depending on the use of stop words (Baeza-Yates, Ribeiro-Neto 1999, 195). Vocabulary and occurrences are often stored in separate files in order to hold the whole vocabulary in memory. Text compression can be used to further reduce the size of the index. For example, the Huffmann and the Ziv-Lempel algorithms have good characteristics for using them on the indexes, e.g., in terms of compression ratio, speed, memory consumption, usability for pattern matching and random access. Another possibility is to build a hash table over the index and to hold only the hash table in memory. The size of the index is relevant not only for search time, but also for construction time. When the index is built, the vocabulary together with the occurrences is held in memory. Once the memory is full, the partial index is written to disk and a new partial index is developed. When all documents are indexed, the partial indexes have to be merged. Two index parts are examined, identical index terms in the vocabulary are merged, their occurrences appended and the new partial index is written to disk. This is done with all index parts of the first generation so that half as many parts exist after the first cycle. The procedure is repeated until only one index is left. Merging can take up a considerable amount of the entire indexing time (about 30%). Query analysis and search. Once the index has been built, users can initiate search processes. The queries are analyzed and searches are executed before results are presented and can be navigated. The types of queries a user can initiate depend heavily on the index, its structure and the search algorithm. The pre-processing that took place during index construction has to be applied to the search terms within the queries as well. The simplest form of search is a single-word query, but often user needs cannot be expressed in a single word. Keywords have to be connected with

4.1. Investigation Services

Boolean operators, i.e. AND, OR and NOT. Usually, search engines use the AND operator by default. As a result, search engines initiate singleword queries for every given keyword and merge the results using set operations. Therefore, Boolean search is classified as a set-theoretic search approach. It is noteworthy that search engines do not use a NOT operator in a set-theoretic sense as this might require time-consuming processing of a probably huge set of documents, i.e. all documents that do not contain the given keyword. Instead, the operator BUT was introduced that operates only on the result sets of the other single-word queries and selects those elements that do not contain the keyword specified in the NOT clause. Boolean searches have one important drawback because they are limited to simply distinguishing between documents fulfilling the Boolean search expression and documents that are not covered by the search expression. Thus, all documents selected are seen as equally fulfilling the criteria so that ranking is impossible in plain Boolean search systems. Fuzzy logic was introduced to overcome this limitation. With fuzzy logic, the result of checking whether a document fulfills a given search expression cannot only be yes or no, i.e. 1 or 0, but also something in between, i.e. floating point numbers between 1 and 0. That opens room for additional search criteria like term frequency and proximity of terms that represent the degree to which a document fulfills the user need. Another feature that many users expect from search engines is pattern matching, because they are used to it at least in a basic form, e.g., from the search for filenames in the operating system. Several types of pattern matching can be distinguished: prefix search, e.g., bus would find bus, busy and business, suffix search, e.g., ting would find interesting and resulting, substring search, e.g., tal would find talking, rental and metallic. Examples for advanced pattern matching techniques are allowing errors and regular expressions which are briefly described in the following. Allowing errors means that misspelled words in queries or in documents, e.g., due to errors in character recognition or authors’ mistakes, are matched with the correct version of the word. The technique is based on similarity between correct word and misspelled version. A common measure for similarity is the edit distance. Edit distance between two words is the minimum number of character insertions, replacements or deletions that are needed to convert one word into the other. If the space character is included in the search then also erroneous split-ups of words can be compensated. Sometimes, this kind of search technique is sold as phonetic search, although phonetic search must be based on more complex linguistic models instead of simple character replacements. An application area for phonetic search is the retrieval of persons based on their names, e.g., to

261

Fuzzy logic

Pattern matching

Allowing errors

262

Regular expressions

Phrase search

Internet search engines

Intranet search engines

4. Knowledge Services

find all persons called “Maier” no matter whether they are written with e or a, i or y. The last type of pattern matching is regular expressions that are usually combined with the use of extended patterns. Basically, regular expressions define three operations on the basis of substring search: union, concatenation and repetition. Union finds words containing either of the substrings like the Boolean OR and is written (s1 | s2). Concatenation finds words that contain the first substrings immediately followed by the second and is written (s1 s2). Repetition finds words with zero or more contiguous occurrences of the substring (s1*). Most implementations of regular expressions also contain extended pattern matching capabilities like wildcards to replace one or more characters, conditional expressions or character classes such as digits, letters, vowels and consonants. Searches for phrases might sometimes return better results than searches for a combination of keywords. This is especially useful if the words contained in the phrase are not helpful as discriminators. Consider the phrase “to be or not to be” which is associated with William Shakespeare and Macbeth. None of the words would be a helpful query term, although the whole phrase will surely find documents dealing with Shakespeare and Macbeth. Searches for phrases demand different indexing strategies and full text search in a narrow sense as stop word elimination and stemming is counter-productive in this scenario. Types of search engines. This section will be concluded by an overview of types of search engines that each address different usage scenarios and also distinct challenges with regard to preparation steps outlined above. Internet search engines offer search services in order to search HTML pages and other widely distributed document formats such as pdf and Microsoft Office document types. Examples for well-known Internet search engines are Google, Yahoo and MS Live Search. It may make sense to search with more than one search engine since the parts of the Word Wide Web indexed are not completely congruent. As this is inconvenient, a number of so-called meta search engines query several search engines automatically, merge the results and present them to the user. Examples for meta search engines are MetaCrawler, Dogpile and Vivisimo. Intranet search engines have the advantage that they can be customized and enhanced for special use cases. This is especially true for the inclusion of resources in proprietary formats. Also, if the structure of the documents in the document set is similar, e.g., conference or journal papers, some meta-data can be extracted automatically and be used to enrich traditional full-text search with meta-data based search techniques. Examples of wellestablished Intranet search engines are Autonomy IDOL, Fast ESP and Endeca Information Access platform, but here as well the big infrastructure vendors compete strongly like IBM with Omnifind, SAP with TREX and Oracle with Secure Enterprise Search.

4.1. Investigation Services

263

Desktop search engines have become more and more relevant in the recent years as it is more and more challenging for many users to structure and retrieve the large amounts of contents filed on their individual PC. This particularly includes emails that still cannot be easily searched through by means of the functions of standard email clients. Although most email clients offer search services, they are often slow and limited to only email contents but do not include email attachments. An important function of desktop search engines is not only to make contents easily retrievable by means of queries that many users know from Intranet search engines, but also to connect the various isolated areas that exist on one single machine. Amongst others, desktop search engines are offered by large providers of Internet search services such as Google, Microsoft and Yahoo. Though they are limited to a local search domain, their general architecture and the steps that need to be undertaken for preparation and index generation are similar to the ones outlined above. There are also a number of innovative search engines. Some of them are specialized in multimedia search, such as finding audio files by humming the melody5, finding pictures based on the rough distribution of colors6 or based on given images (query by example7) or finding 3D models using a number of 2D drawings8. Other search engines specialize in searching product information instead of the whole Web9, present points of interest geographically in a map10 or search in a certain domain (e.g., scientific publications11). There are also some promising developments regarding clustering12 or graphical formatting of results13. In large-scale organizations, often more than one Intranet search engine software is in place, e.g., SAP TREX for documents stored in the SAP system, Microsoft Sharepoint Search for documents inside Sharepoint and a Google search appliance for indexing file servers. Instead of indexing every document three times it would make more sense to agree on a common indexing standard and reuse the index created by another search engine. OpenSearch14 is such a standard which allows e.g., to reuse indices from Coveo Enterprise Search and EMC Documentum Search Service in Microsoft Search Server 2008. Similarly, Microsoft desktop search can integrate search results from MS Search Server.

Desktop search engines

5 6 7 8 9 10 11 12 13 14

URL: http://www.musicline.de/de/melodiesuche, last access: 2008-11-01 URL: http://wwwqbic.almaden.ibm.com, last access: 2008-11-01 URL: http://viper.unige.ch/demo, last access: 2008-11-01 URL: http://shape.cs.princeton.edu/search.html, last access: 2008-11-01 URL: http://froogle.google.com, last access: 2008-11-01 URL: http://maps.google.com, last access: 2008-11-01 URL: http://scholar.google.com, last access: 2008-11-01 URL: http://clusty.com, last access: 2008-11-01 URL: http://www.kartoo.com, last access: 2008-11-01 URL: http://www.opensearch.org, last access: 2008-11-01

Specialized search engines

Federated search

264

4. Knowledge Services

Semantic search engines

A promising direction for the enhancement of search results is the application of ontologies. Today’s search engines are not able to understand the semantics of a search query. Particularly, they do not identify the types of concepts that a user is searching for, e.g., people, places and products, but rather they rely on the exact terms specified within a search query. Ontology-based search engines relate parts of the search query to semantic concepts and thus are able, e.g., to offer additional results identified by inference mechanisms or to filter items that are not relevant. For example, if a user searches for the keywords hybrid and car using a “traditional” search engine, he receives a list of resources that contain these keywords. An ontology-based search engine may identify related concepts such as producers, sellers, specifications and locations, offer additional information structured according to these dimensions and guide the user through later steps of the search process. For example, she may subsequently focus on producers and specifications if she is interested in technological details or she may add a location dimension and concentrate on concrete products available within a specific region. However, one of the preconditions of ontology-based search is that the resources to be indexed are semantically annotated, e.g., based on RDF and OWL (see section 3.3.3). Consequently, semantic search in the near future will be mainly restricted to defined domains such as organizational Intranets and will not be generally available on the Internet.

4.1.4 Presentation Figure 4-8 displays the steps to be supported by presentation services. search services

presentation services

1. acquisition & pre-processing

4. ranking index 5. result visualization

2. indexing 3. query analysis specify/refine query

Figure 4-8.

user profile

6. feedback

navigation of results

Steps supported by presentation services

They continue the search steps as described at the beginning of the previous section. After query results have been identified based on the index,

4.1. Investigation Services

265

they need to be ranked according to their relevance. They are then visualized in a way that acknowledges user-specific needs. The user profile included within the figure may support this as it contains information about, e.g., user interests, competencies and learning goals. Feedback includes the gathering of information about the actual relevance of the results for the user as well as support for the reformulation of the initial search query, i.e., the search process may be executed multiple times. Ranking. With ranking, search results are sorted in a way that suits the user needs. Items that most probably fulfill users need are ranked higher than those that are expected to suit the user needs only to a lower degree. This is determined by comparing items based on a set of selected criteria and with the help of a mathematical model. Besides set-theoretic search models that were briefly introduced above, there are also algebraic and probabilistic models. The most prominent algebraic method is the vector model. Both, query and result document are represented as n-dimensional vectors of weighted terms. The similarity of these two vectors is then calculated. A measure for similarity is the mathematical comparison of them that can be expressed e.g., in the cosine of the angle between the vectors. Artificial neural networks are another technique that is used for pattern matching and ranking in search engines. Neural networks try to imitate the structure of human brains by using artificial neurons that are interconnected with weighted connections. Query terms thereby act as input, neurons consist of index terms and documents correspond to the output. Weights at index terms must be adjusted to get good results. The network is usually trained before it is used for searching, but can also be further improved using user relevance feedback to adjust weights during normal usage. Probabilistic models calculate the probability that a set of documents matches the user need. It would be no problem to retrieve the relevant documents if the system would know the exact properties that describe the document set. However, the user query is only an incomplete and maybe even erroneous representation of user need. Therefore, assumptions about the properties of the document set must be made that can be improved step by step using user relevance feedback. The user has to specify whether the retrieved documents are good or bad matches. The result is refined with the help of this additional information. A popular implementation for a probabilistic search method is a Bayesian network. Bayesian networks are directed acyclic graphs in which nodes are random variables. Arcs represent causal relationships between these variables. Strengths of these relationships are expressed by conditional probabilities (Baeza-Yates, RibeiroNeto 1999, p 48f). The advantage of Bayesian networks is that different ranking criteria like past queries, feedback and current query terms can be combined using a clear formalism.

Search models

Neural networks

Probabilistic models

266

4. Knowledge Services

Criteria for ranking

The key challenge for ranking is finding good criteria that allow to determine to what extend documents fulfill the user need. Some of these criteria are briefly described in the following paragraphs. Often, a combination of several criteria yields best ranking results. One criterion is term frequency. Documents that contain query terms more often are expected to be more relevant to the user. If no exact matching is being required, similarity of query terms and occurrences can also be a criterion. Exact matches are usually better than matches as substring or matches with errors. Proximity of the occurrence of different query terms is also a criterion. The closer together the terms occur (e.g., within a sentence or paragraph), the more relevant the document can be expected to be. Besides these basic techniques, some enhancements to ranking algorithms are discussed. One possibility is to compare data about the document’s author and the searching user to determine better or worse matches. It can be argued that a domain expert regularly (but not always) might not want to retrieve documents that contain an introduction to the domain she has already mastered. Vice versa, learning theory suggests that novices might learn more effectively from (documents produced by) intermediates compared to experts in the regarding domain. Thus, if skill profiles of both, user and author, are available, they can be compared to rank the results. Other criteria like document type (conference paper vs. presentation), reputation of authors or institution the authors work for (e.g., a renowned research institute vs. private, anonymous opinion contributed to a newsgroup), creation date (recent vs. out-dated) or access frequency of the documents (used daily vs. never retrieved since creation) can be used if the corresponding data is available and the use case permits to do so.

Recent developments in result ranking

Result visualization. Once the results are ranked, it has to be determined which information exactly should be presented to the user, at least this is the filename and location of the resource. Additional information helps the user to decide whether the document should be inspected in more detail or can be discarded. Often, a percentage is presented that shows relevance of the item for the given search term calculated by the search engine. Although often differing a lot from the user’s subjective rating, it still helps for orientation, e.g., if there is a great gap between the first number of results and the following ones. A summary of the document contents, especially the parts in which the query terms occur, can provide additional information that supports quick evaluation of the results. Creation date, size of the document in byte or even better in words or pages, authors, as well as time of last access or number of times accessed since creation are other examples of useful meta-data that should be presented together with the results. Meta-data primarily describes the context of creation, but could also be extended to cover the context of application, e.g., the skill level targeted or other descriptions of intended user groups that might be evaluated by the search system and then presented to the user. It is also

4.1. Investigation Services

267

important to not only deliver documents as results, but include information about domain experts, email messages and community home pages that are relevant. Finally, the result list could include information about related documents identified based on a collaborative filtering approach such as “users who viewed this item also viewed XYZ” (section 4.1.2), or the cluster, group or domain a document belongs to. Feedback and iteration. An optional step after result presentation is the refinement of the search. This can be accomplished either by reformulating the query, restraining the result set or by providing relevance feedback to the search engine. Reformulation leaves improving the relevance of retrieved results completely to the user who has to find better query terms, often based on words found in the retrieved documents. Especially the exclusion of words that blur the results can increase result quality. This technique requires competent users and at least a basic understanding of the underlying search algorithms. Restraining the result set is often linked to categorization or clustering of results. Users can determine which group of documents suits best their needs and limit results to those that are within this group. This is effective, if there is either a good clustering algorithm or a good manual classification of documents. This kind of feedback is especially interesting if one or more of the query terms have different meanings in different contexts. A technique that leaves improvement of results completely to the search engine is relevance feedback. The user only specifies which documents are close to his or her need and which are not. The search engine then analyzes the documents that were rated by the user and tries to extract discriminating terms, i.e. the properties that differentiate useful from useless results. Based on these terms, a new search is being initiated and new results are presented to the user. If there are documents in the document base that match the user's needs, then usually three iterations are sufficient to reach a good result set. This method is based on probabilistic search algorithms and is well suited for inexperienced users or non-text documents (see multimedia search in section 4.1.3). The feedback can also be stored in a user profile which can be used for further searches to enhance the initial assumptions about user needs.

4.1.5 Monitoring Monitoring services support the continuous survey of defined information sources. In contrast to search and specifically exploration, the user already knows about the resources and topics that he wants to stay informed about, which may be determined by concrete work tasks, interests or learning goals. Important monitoring services are syndication mechanisms and information agents which will be described in the following.

Reformulation

Restrainment to a cluster

Relevance feedback

268

4. Knowledge Services

Syndication

Web sites do not always contain contents that are generated by the Web site owner but may also include information that is provided by so-called content providers, e.g., news items, stock indices and weather forecasts. For such sets of contents that are created and published for distribution and reuse, the term syndication is applied (Boiko 2002, 86f). The customer of a syndication service usually includes statements into the Web site that request pre-defined content elements from a syndication service offered by the content provider. Examples for widely-spread XML-based markup languages are the Really Simple Syndication standard (RSS)15, NewsML and ICE (Information and Content Exchange). Before the content elements are published, they are transformed into the right presentation format by means of templates. During the last years, the RSS format has been increasingly used not only for “traditional” syndication but also for the individual aggregation of information published on the Internet. More recently, the Atom Syndication Format has been suggested in order to integrate various suggestions that enhance the RSS standard and particularly in order to acknowledge the need to include various different types of contents including audio, video and downloadable program files into syndication feeds. Feedreaders, also called aggregators or RSS-readers, can be applied in order to subscribe and display RSS news feeds. These are offered not only by news Web sites but increasingly by various kinds of Web resources with frequently updated contents, e.g., Web logs and discussion forums. A feedreader thus enables individuals to conveniently keep track of various different sources of information by means of one integrated interface. Many software programs used for email access or Internet browsing have RSS feedreaders integrated in their recent versions, e.g., Opera, Mozilla Firefox and Microsoft Internet Explorer. The RSS standard is also used for the distribution audio and video files which is then referred to as podcasts. The term combines the words iPod, i.e. a MP3 player of Apple, and broadcast, i.e. the distribution of audio and video to a larger audience. The distribution of audio and video files by means of RSS feeds enables users to easily monitor their preferred Internet sources. New items are downloaded automatically may be conveniently transferred to mobile appliances such as audio and video players. Today, it can be chosen from a large variety of podcasts that usually can be subscribed without costs. Listing 4-1 shows an example for a newsfeed based on the RSS 2.0 specification. It also contains additional XML elements that adhere to the podcast DTD of Apple’s iTunes which are marked by a corresponding namespace.

News aggregation with RSS

Podcasts

RSS example

15

The acronym RSS in earlier versions than the current specification 2.0.1 (URL: http://www.rssboard.org/rss-specification, last access: 2008-02-28) is used for Rich Site Summary and RDF Site Summary.

4.1. Investigation Services

InstantNews Today Articles of InstantNews http://www.inews.com/ Wed, 03 Sept 2008 15:25:55 +0100 Inhouse CMS 1.0.2 http://www.inews.com/inews-logo.gif InstantNews Logo http://www.instant-news.com/ InstantNews Feed. Click to visit. Editors inews.com [email protected] Info Instant News items today InstantNews InstantNews - inews.com en 60 China’s economony is expected to grow slower http://www.inews.com Economists state that China’s industry may cool down [email protected] Wed, 03 Sept 2008 13:17:52 +0100 http://podcast.inews.com/ef9.mp3 Burg, Tony 00:03:24 ... Listing 4-1. Example of RSS podcast newsfeed

It contains two main elements: channel and item. Channel represents the overall podcast and is characterized by a number of sub-elements

269

270

4. Knowledge Services

that contain information concerning title, description, URL, language and copyright, time of last update in the element lastbuilddate, the software used for creating the feed in the element generator, an image characterized by further sub-elements, i.e. url, title, link and description, a category for the contents, and the number of min-

Personal information agents

utes that a channel can be cached before updating from the source enclosed by ttl which stands for time to live. iTunes-specific elements are name and email of the owner, an image to be displayed in the iTunes software as well as subtitle and author of the podcast. One channel usually contains multiple items that are further specified by title, link, description, author and pubDate. Whereas link may contain the general URL of the item, enclosure points to location of the media object referenced by the item. Guid is a string that uniquely identifies the item, which in the example also includes iTunes-specific elements such as author and duration of the media file. Please note that not all elements specified in the RSS standard are included within the listing. It also refers to a stylesheet podcast.xsl that can be used in order to display the feed within a Web browser that implements no specific support for RSS. Personal information agents help users to monitor defined resources by notifying them about associated events such as new elements or changes to existing items. Though the term “agent” suggests an autonomous and active entity, information agents are regularly designed in a much simpler way, i.e. as a mechanism that frequently issues the same search query and then analyses differences of the current and of prior search results delivered by a search engine. This may also include changes on actual contents such as Web pages located somewhere on the Internet. Users may get notified about changes by email or may access the agent’s interface for information such as search and monitoring statistics. The Copernic Agent family of software products16 is an example for this software class. Figure 4-9 shows an example for an Intranet information agent in Open Text Livelink. The user may navigate to the configuration page shown in the lower right by selecting “set notification” on any folder of the Intranet. Then, she may configure a set of events that she wants to get notified about, e.g., if new items are added. These events are listed by the notification agent within a report determined by the user. Reports summarize all relevant events associated to the folder configured or within one of its subfolders. They are distributed by email at specified time intervals. As a consequence, users are able to stay informed about the activities in selected Intranet areas without having to continuously access and manually determine the changes done by other users.

16

URL: http://www.copernic.com/en/products/agent/, last access: 2008-03-04

4.2. Individuation Services

Figure 4-9.

Example Livelink notification mechanism

4.2 Individuation Services As stated in section 1.4.4, 42ff, the concept of individuation has so far been neglected in many initiatives to create knowledge infrastructures. This means that many services in this category are so far quite rudimentary and consequently it is not surprising that this section is considerably shorter than the other sections in this chapter, because the other services have been deployed systematically for a much longer period of time. Many efforts have focused on transparency of knowledge and on supporting knowledge workers in sharing knowledge or even detaching knowledge from humans as “media” of knowledge. However, at least in a more individualistic culture stressing diversity, the individual knowledge worker requires also support for her activities concerning the appropriation of knowledge towards the individual subject. This includes (1) the reflection of individual experiences, (2) the expression of ideas and proposals in a way that makes sense for the individual, (3) building competence and managing one’s individual career and (4) achieving the final step of individual professional development, expertise in a certain knowledge domain. Figure 4-10 gives an overview of four classes of EKI services helpful for individuation.

271

272

4. Knowledge Services

experience

expertise

expression

competence

Figure 4-10. Overview of individuation services

Due to its newness, in the following the term is first defined in section 4.2.1 before the corresponding services are described in sections 4.2.2 to 4.2.5.

4.2.1 Foundation Concept of individuation

Experience

Expression

Competence

The concept of individuation is widely used in a number of scientific disciplines, most notably philosophy and psychology. Its origin can be traced back to the Latin adjective “individuus” which means indivisible, inseparable or undivided. It describes processes in which the undifferentiated becomes individual, or processes in which separable components become an indivisible whole. When translating this rather abstract definition to EKIs, it comprises four important steps in the maturing of knowledge that can be supported by services. First, knowledge workers have to make sense of a vast and chaotic amount of material, e.g., the material available on a company Intranet or the WWW. Thus, they shift their focus or awareness to knowledge elements which they differentiate out of the mass of material and connect to them. In the most simple way this might mean reading and understanding a resource on the Web and then tagging or bookmarking it in order to connect to it. Second, the knowledge worker then needs to make sense out of the separate knowledge elements that she has connected to in order to learn and, in subsequent steps, to build competence and expertise which means the encapsulation of knowledge elements into an inadvisable whole. In the most simple way this might mean connecting contents from diverse sources in order to express what the knowledge worker knows about a certain knowledge domain. It might also mean that the knowledge worker expresses some personal idea or proposal that is at first unconnected to the rest of the knowledge elements accessible to the knowledge worker. While the services for these two steps are quite straightforward, supporting individual competence development is much harder to achieve. Competence reflects the relationship between an individual’s skills and the

4.2. Individuation Services

requirements of the work to be competed by the individual. Although competence services are related to the KM instrument competence management, they aim at supporting individual knowledge workers to develop their own knowledge and skills in a self-guided way. This includes reflecting on and making sense out of the development of individual competencies in the pursuit of a sequence of tasks performed in a business or organization. What can be said about services for competence development is all the more true for services targeting development of expertise. While competence reflects on the fact that an individual’s skills are sufficient to complete work in a certain domain, expertise reflects on an individual knowledge worker’s long-standing experience in a domain of knowledge which differentiates her perception and acting in the domain from non-experts. Some authors extend the hierarchical model of data - information - knowledge (section 1.1.1, 4ff) by competence and further on by expertise or wisdom which should reflect the increasing abstraction from the concrete happenings, but also the increasing integration of separate individual experiences into a coherent whole - an expert’s profound knowledge that marks the highest (supportable) step of individuation with respect to knowledge workers’ professional development. In the following, these four steps are used to structure the discussion of services for individuation in an EKI.

273

Expertise

4.2.2 Experience Experience is the foundation for building professional competence and is a typical characteristic of individuals. From the perspective of knowledge workers, a substantial part of experiences reflects the handling of information and this is where the services of an EKI can help. The implementation of experience management services from the perspective of individuation basically eases documentation, integration and retrieval of personal experiences. These services have to be integrated into the daily work practices of knowledge workers in order to be accepted. Technically, this can be supported by notes services. Basic support for management of personal experiences, but also for keeping ideas is given by the notes functionality often included in personal information management or Groupware systems. Although the functionality is quite simple, it is helpful since notes are stored safely on the server, can be taken directly within a program that is opened all day anyway (for receiving emails) and can be replicated easily to a PDA or Smartphone (section 5.3.3, 399ff) and therefore are always accessible. Microsoft OneNote, Evernote, Jarnal are examples for note-taking software. Depending on personal favours and organizational culture, word processors or email clients might be used as well. Finally, many ideas are simply discussed with colleagues during lunch break or after a meeting

Experience management

Notes

Note-taking software

274

Thought structuring and problem solving

Offline reading and downloads

Tagging

Tags, posts

4. Knowledge Services

with no electronic support at all. In any case, the results are not easily integrated into the EKI and therefore not available to others. Methods for structuring of thoughts and for supporting individual and/or group problem solving have a long tradition in research and practice. An example for a widely used thought structuring software is Mindmanager, a personal mindmapping tool. However, there is an increasing number of online and collaborative note taking tools, such as Mindmeister, an example for an online collaborative mindmapping tool. This points towards an important requirement in the seamless integration of services along the maturing model. In other words, the same services are used for expression of individual experiences or ideas as well as for their subsequent sharing and integration with experiences or ideas of others using interaction services (see section 4.3, 285ff). Further examples for methods and instruments to support individual problem solving are SQ3R, information mapping, concept cards, micro-articles, concept mapping or story templates (Reinmann/Eppler 2007). As long as the telecommunication infrastructure is not as ubiquitously economically feasibly available, most knowledge workers still sometimes need to work offline and thus require functions for managing downloads. Web browsers such as Mozilla Firefox offer plug-ins, e.g., ScrapBook, for increasing effectiveness of reading Web pages offline and managing one’s downloads of HTML pages, snippets and entire sites from the Web. Document management systems such as Open Text Livelink offer integration into the MS Windows Explorer including offline functionality that helps in synchronizing files changed offline with their online counterparts. Adding annotations to Web resources is called tagging. Thereby, a tag is a comment unrestricted to a controlled vocabulary and represented in different kinds of structures, which have come to be called folksonomies because they are collaboratively developed (see section 4.3.5, 301ff, collaborative tagging). Unrestricted to a controlled vocabulary means that the individual knowledge worker is free to choose whatever keyword she wants to assign to a resource. The basic element of information in such a system is the following triple: tag = user resource keyword Frequently, a user assigns multiple keywords to the same resource in a single step which is called a post: post(user n) = (timestamp, tags

tags(user n resource r keyword ))

Tags can be used by knowledge workers in order to organize the personal workspace and for searching and browsing. Easy handling can reduce the time required for tagging, the barrier to contribute and the costs for the organization. Knowledge workers typically find it difficult to contribute with established methods for structuring resources using controlled

4.2. Individuation Services

vocabularies, however, because structuring thoughts in a given format represents a significant barrier to participate. Also, controlled vocabularies primarily used in libraries are not always transferable to the wide variety of digital resources. Tagging with its unrestricted opportunities can reduce these barriers and motivate users to contribute. This model has been successfully applied on the Internet. Typically, users can add free-text descriptions to Web resources and additionally they can select descriptions from a list of the most frequently used keywords, for example in delicious.com or connotea.org. In many organizational applications such as document management systems (see section 4.4, 305ff), the author of a document is required to fill out a number of metadata fields, e.g., in order to categorize a document according to its type, related organizational units, its focus and application area or just to assign descriptive text and keywords to the item in question. General goal of this task is to make retrieval of knowledge elements easier. However, the author may not always be able to anticipate relevant classifications and keywords that a potential user would associate with it during the discovery process. Vice versa, if users bookmark resources for later retrieval, they engage in a very similar process of content classification. With respect to a useful application in organizations, the task of annotating knowledge elements could be moved from professional meta-data editors to all knowledge workers who are about to appropriate a certain knowledge element to them. The basic idea of collaborative tagging is to offer means for an easy assignment and exchange of keywords that can be freely chosen by authors and users. Those tags are concurrently used as bookmarks which lead to the synonymous term social bookmarking. Tags and posts can be statistically analyzed and lead to tag clouds, tag networks or tag clusters which in turn can improve the workings of integration services (see chapter 3, 151ff). The system of keywords that evolves step-by-step is also called folksonomy, a combination of the words “folks” and “ontology”, because the tags represent a network of terms that has some basic similarities to ontologies (see section 3.3, 205ff). However, they are not built on a defined semantic basis and therefore offer limited possibilities for advanced reasoning and inference. In order to be of value, folksonomies have to converge at some point, i.e. relevant and important keywords need to be visible to users in terms of a high frequency of occurrence. However, the process of convergence is often encumbered or even hindered by spelling errors, variations of terms such as singular vs. plural forms and multiple languages. Figure 4-11 denotes the basic triple of user, resource and keyword that a tag is comprised of.

275

An Internet success model for use in organizations

Collaborative tagging, social bookmarking

Folksonomy

Tagging process

276

4. Knowledge Services

KW1

KW1 KW2

KW2KW3

KW4 KW5 KW6 KW7 KW8 KW9 KW10 KW11 KW12

KW8 KW1

KW8 KW4

KW1

KW2 KW6

KW4 KW2

Figure 4-11. Collaborative tagging process (legend: KW=keyword)

The keywords assigned to a resource can be used by an individual for later retrieval of that resource (1). However, benefits increase largely if many users annotate a single resource with keywords. This renders some keywords more important than others in retrieval due to the fact that more users have used that keyword. This is typically visualized by varying font size and emphasis (2). An individual knowledge worker, a subject matter expert or a knowledge administrator might then be responsible for structuring keywords, e.g., into a simple tree structure (3) in order to further aid retrieval of documents, visualized by a path from KW8 over KW4 and KW2 in (4).

4.2.3 Expression Concept of expression

Difference to in-form-ation

The term expression can be traced back to the Latin word exprimo, expressi, expressus which means to press, squeeze or bring out something, to articulate or to describe. Expression is a term that resonates well with the motive of self-realization in Maslow’s hierarchy of needs. It refers here to bringing out thoughts and ideas, to make them explicit, to explicate them or, in Nonaka’s terms, to externalize them (section 1.1.2, 15ff). This sounds quite similar to the activities which are targeted by inform-ation services (section 4.4, 305ff). However, expression as part of individuation only follows the needs, requirements and formalities of the individual knowledge worker and is not (yet) put into a form officially legitimated by the organization the individual works for. This might be a protocol or a formal report stored in a DMS with privileges assigned by a formal representative authorized to do so. In the case of expression, it is the individual knowledge worker who decides on who might see the contents and thus manages access privileges.

4.2. Individuation Services

277

Services to support this model have been created under the umbrella of peer-to-peer systems (see ”Peer-to-peer architecture” on page 51) which bridge the gap between individuation and interaction services, e.g., in MS Groove (section 4.3.5, 301ff). Moreover, there are also systems used to bridge the gap between personal, informal individuation services and organizational, formal in-form-ation services. An example is MS Sharepoint Server that offers the knowledge worker a personal area for creating and handling individual resources that can be transferred later on into shared resources (interaction) and even formal resources part of legitimated containers (in-form-ation). The provision of corresponding services needs to take into account that curiosity and creativity are the main drivers behind the desired activities of knowledge workers expressing new ideas and insights and feed them into organizational innovation processes. Although these are processes which need talent and a fostering organizational culture that boosts them to a large extent, there are some services that can support these processes by removing barriers that hinder knowledge workers from explicating ideas, or knowledge. Web logs or blogs are tools that recently gain a lot of attention for easy expression of personal experiences and making them accessible to others. Blog as a term is a contraction of Web and log and denotes a form of content management system typically operated by a single individual. Blogs technically work similar to WCMS with a rigid structure and are organized using the dimensions time and topic. New Blog entries, also called posts, can be created easily using a standard Web browser. Blog posts are typically presented in reverse-chronological order and as opposed to more closed forms of Web content management, the idea is that Blog posts do not remain as stand-alone entities, but are extended by capabilities to e.g., search: Blog posts can be indexed and offer keyword search on this basis so that corresponding articles can be accessed by colleagues, link: Blog posts can be connected to posts in other Blogs extending the usual linking capabilities that HTML offers, syndicate: the content can be reused with the help of RSS (RDF Site Summary) feeds that can be integrated into the HTML code of other Web sites and deliver abstracts of new entries in Blogs which can also be pushed to knowledge workers e.g., via email, comment: Blogs offer functions that allow other Bloggers to easily hand on feedback directly into the Blog. Blogs have become renowned for the currency of their entries, additions and changes independent of the current physical location of the Blogger and personalization of content management in organizations. By commenting and referencing other Web logs, a kind of community emerges. Such a community is called the Blogosphere. Summing up, it seems that Blogs can be used for expressing personal ideas or experiences, comment-

Bridging individuation, interaction and in-form-ation

Expression services

Blog

278

Microblog

Experience management system

4. Knowledge Services

ing on collections of snippets from the Web, appropriating them, i.e. connecting them with oneself and making this visible to a community, the entire organization or the public Internet. Blogs might be useful to close the gap between well-structured experience documents handled in DMS or file servers on the one hand and notes with ideas in personal information management or Groupware systems on the other hand. Microblogging is a sub-form of blogging in which posts are restricted to brief texts of up to 160 characters that cannot only be submitted from and received by a Web browser on a desktop or laptop PC connected to the Internet, but also e.g., by an email or instant messaging client or by a mobile phone using short messaging service. The multimedia version of this relies on multimedia messaging service (MMS) that are translated into Blog entries including the picture sent to the Blog server. Ideas and the reflected essence of personal experiences can also be fed into experience management systems. As stated before, these systems have to be integrated into the daily work practices of employees in order to be accepted. Within organizations, knowledge workers might find it cumbersome to express the context of an idea or experience they want to appropriate to themselves. A tool might help in easily attaching context dimensions to content expressed by the knowledge worker during work processes. One example for such a solution is the toolbar developed in the APOSDLE project17.

4.2.4 Competence Concept of competence

The term competence is used differently in e.g., psychology, pedagogy and management science. In the latter, the concept can be bound to a group, an organizational unit or even an entire organization, i.e. an organizational competence consisting of a bundle of compatible resources (see section 1.1.1, 4ff), to a job position, i.e. competence in the sense of authorized to perform a task (“may do”) or to an individual employee, i.e. competence in the sense of capable to perform a task (“can do”). Here, the term competence has a relational connotation between an individual’s knowledge, cognitive, social and behavioral skills on the one hand and the goals to be achieved, the requirements job or task to be done on the other hand. Some distinguish between the term qualification which has a strong connotation of a profession, role or task and the term competence which is not as narrowly bound to a profession or task. Thus, competence denotes a capacity to interact with the environment or, more specifically and from an organizational perspective, a capacity to act with respect to a certain domain or task at hand. This means that the requirements of a task can be fulfilled by a competent individual.

17

URL: http://www.aposdle.tugraz.at/, last access at 2008-11-01

4.2. Individuation Services

Competencies are typically classified into (1) professional or subjectoriented, (2) method and (3) social-communicative competencies. Examples for subject-oriented competencies are technical competencies, business knowledge or knowledge related to a specific industry sector. Examples for method competencies are management methods, problem solving, project management, structuring or analytical skills. Examples for socialcommunicative competencies are communicative skills, rhetorics, ability to work in teams or the capability to convince people. Knowledge and skills are acquired to a substantial extent in some formal education prior to the first job, e.g., school, university, but many are refined, developed or even acquired in trainings into the job, off the job or on the job. An active, self-guided role of an individual knowledge worker in reflecting on the complex interlacings of her competencies might be supported by competence services. This type of services provides support for the personal part of the KM instrument competence management. This means that an individual knowledge worker is supported when systematically monitoring, guiding and reflecting on the development of her functional-professional, method and social or communicative competencies. Thus, the services coarsely can be divided into two groups: services for monitoring or visualizing competencies and services to support the development of competencies. These will be described in the following. Monitoring one’s competencies potentially involves a large number of diverse data that has to be brought together to give an indication on what state of competency the individual is in. Examples are educational background, participation in projects, participation in trainings, certificates and participation in formal career development programs. Personal competence management builds upon a framework of competencies that can be used for profiling by the individual knowledge worker. Frameworks are developed and maintained by e.g., professional organizations. Examples are the framework of competencies required for ICT professions managed by Swiss ICT18 or the Skills Framework for the Information Age (SFIA) managed by the SFIA foundation19, backed by e.g., the British Computer Society. Competence services thus build upon such a framework and allow for self-assessment of the individual according to the competencies listed in the framework. The individual typically can contrast a target profile with an as-is profile and see what measures could take her from here to there. Human resource management might also offer services that allow the knowledge worker to compare her competencies with those competencies required by current job openings or by future probable job openings in the same organization. Although monitoring is personal and first-hand only accessible by the individual, it might also be released and accessible

18 19

URL: http://www.swissict.ch, last access: 2008-11-07 URL: http://www.sfia.org.uk, last access: 2008-11-07

279 Professional, method and social competencies

Competence services

Monitoring and visualizing competencies

280

Knowledge asset map

4. Knowledge Services

by colleagues within a team or community, a supervisor, coach or a HRM employee in a protected area, or public to the entire organization. Knowledge maps can be used to visualize competencies made accessible by individuals not only for themselves, but also for teams of knowledge workers and organizational units. Figure 4-12 maps (1) central areas of competence in an IT consulting organization and (2) employees according to (3) their level of expertise.

Figure 4-12. Example for a knowledge asset map (Eppler 2008, 134)

Developing competencies

The squares indicate whether an employee holds basic knowledge, expert knowledge or is a leader in the corresponding area of competence. The map shows the importance of Mr. Tinner and Mr. Ehrler for the organization because they seem to be competent in (almost) all relevant areas of competence. Another group of services supports knowledge workers when reflecting on their development of competencies. This involves the development-oriented methods and instruments suggested for personal knowledge management (Reinmann/Eppler 2007). Examples are contact network: a visualization of a collection of the most important internal and external contact persons in defined areas of competency of a person, competence agenda: is a table-oriented compact plan that visualizes work place activities that should help to develop competencies in several defined areas, competence portfolio: is a collection of portfolios that allow a knowledge worker to assess the current state of competencies and develop learning strategies in order to achieve a more favorable position in the

4.2. Individuation Services

281

portfolios. Examples are dimensions that map one’s competency edge over competitors, the degree of application of the competencies, the half-life of competencies or the breadth of application domains for the competencies, lifeline: a competency-oriented overview of a knowledge worker’s past, present and future, synergy map: supports the analysis of how personal (learning or knowledge-oriented) goals are related, whether they are conflicting, neutral or complementary. More recently, increasing attention has also been paid to learning activities for self-guided competence development. “Self-guided” implies that knowledge workers take on individual responsibility for these activities and the learning services that are needed to support them. So-called personal learning environments allow for an individual drawing together of more informal learning services at the workplace or during leisure as well as more formal instruction services provided by an organization the knowledge worker is employed in (see section 4.5, 331ff) or by educational institutions such as Universities. Like personal information management, personal learning environments are the sole responsibility of the knowledge worker and thus their contents and the “mash up” of services, to use a Web 2.0 phrase, are controlled individually rather than organizationally and thus are a central example for individuation services.

Personal learning environment

4.2.5 Expertise As stated in section 1.1.1, 4ff, the concept of expertise targets the individual rather than the organizational actor. There is a long-standing tradition of research in the domain of expertise which focuses on the differences between novices, intermediates and experts with respect to problem solving. The differences carry implications for handling of knowledge and supporting services. For example, whereas novices require external support in selecting and reflecting learning material, experts are supposed to regulate their learning behavior by themselves. With respect to EKI, this affects presentation of knowledge, e.g., novices are presented abstract, generalized fact knowledge whereas experts are presented concrete, individual case examples, meta-knowledge, e.g., about the level of expertise of other participants which are suggested as “knowing” to a novice versus an expert, services for experts that help them to perform self-guided learning (see also section 4.5, 331ff). There is a model well-received in the literature which describes the development of expertise as applied to unstructured situations for which there is no set of facts and factors which fully determine problem, possible actions and goal of the activity, e.g., patient care, business forecasts, social

Concept of expertise

From novices to experts

282

Novice

Advanced beginner

Competent

4. Knowledge Services

interactions. It stresses the importance of implicit knowledge for expert problem solving. In the step-wise course of becoming an expert thinking is reorganized qualitatively which means that expert knowledge is organized differently from explicit knowledge about facts and rules. Teaching means to subsequently lead the learning person from an analytic via a planning to an intuitive way of problem solving. A central concept is “power of judgement” as a holistic way of pattern recognition which is highly adapted to contexts. Adapting a person’s organization of knowledge means replacing knowledge about facts and rules with practical cases. The cases are used as patterns to intuitively judge the adequate actions required in a specific situation. The model distinguishes five steps that are briefly described in the following (Dreyfus/Dreyfus 1986, 19ff): When observing an expert, novices are overwhelmed by the complexity of a situation so that they are unable to imitate an expert. In the first stage of learning, novices are provided with non-situational or context-free attributes and rules. These do not reflect the total situation, ignore the total context and do not require the novice to understand the total structure of the situation. The novice analyzes a situation by spotting single attributes and selects actions according to the rules remembered. The attributes are not implicitly integrated, but explicitly focused and summed up. The advanced beginner has extensive practical experience in the domain. Thus, more context-free attributes can be used in judging the situation and more complex rules determine actions. The most important difference to the novice’s problem solving is the use of so-called aspects. These are situational or context-specific attributes that the advanced beginner has encountered in a greater number of “similar” practical cases. The selection of actions is not only based on context-free rules, but also on context-specific guidelines. However, problem solving is still not integrated as there is no conscious examination of configurations of attributes. The single attributes and aspects are considered as being of equal value. The number of attributes and aspects considered increase to a point where the learner is confronted with an overwhelming number of elements. The central skill differentiating the competent from the two levels before is the potential to analyze a situation with the help of a perspective. The person is able to plan consciously and thoughtfully. Goals and plans increase the complexity of the analysis, but reduce the complexity of the situation because not all attributes and aspects have to be considered anymore. Conscious, analytical problem solving is maximized on this level of expertise. Actions are selected with the help of a chosen perspective. As a consequence of the subjective selection of a plan, competent employees will feel responsible for their actions (emotional involvement). On the two levels before, actions were taken by strictly applying rules and guidelines. Thus, unwanted results could be attributed to inadequate rules or guidelines.

4.2. Individuation Services

The central new skill of the skillful master is the ability to perceive situations as a whole as opposed to observing single attributes and aspects of a situation. This means a holistic recognition of similarities of the current situation with situations the master encountered before. The master has a “mental library” of typical situations perceived using a specific perspective. New situations are perceived from a specific perspective without consciously selecting it. The relative importance of attributes and aspects in the problem domain is not analyzed consciously anymore. The situation rather presents itself accentuated to the master who intuitively expects which situations could follow the current situation. Actions are still selected consciously on the basis of maxims, heuristic principles relating an action to a configuration of attributes and aspects. The master consciously selects actions that have proven successful in a similar situation. Every specific situation the expert encounters will automatically trigger intuitively appropriate action(s). Experts not only store perspective-based types of situations, but associations of types of situations with corresponding actions. Situations are grouped so that they require the same decisions and actions. They are stored in such a number that they cannot be verbally described. Thus, the expert does not process atomic facts logically, but perceives holistic similarities between the current situation and situations encountered before without having to study isolated elements. Conscious analytic planning does not occur anymore. The expert can handle situations in a goal-oriented way, but without consciously setting goals. Table 4-3 shows the five levels of the model with those elements of problem-solving highlighted that determine central shifts between stages. Table 4-3.

283 Skillful master

Expert

Model of the acquisition of expertisea

skill level

components

perspective decision

commitment

1. novice

context-free

none

analytical detached

2. advanced beginner

context-free and situational

none

analytical detached

3. competent

context-free and situational

chosen

analytical detached understanding/ deciding; involved in outcome

4. proficient/ context-free and skillful master situational

experienced analytical involved understanding; detached deciding

5. expert

experienced intuitive

context-free and situational

involved

a. Dreyfus/Dreyfus (1986, 50)

Experts not only have more profound area-specific knowledge, but also apply so-called schemes to analyze situations which allow them to consider more information quicker than novices. Experts are also quicker in

Experts and novices compared

284

Consequences for knowledge infrastructures

Expertise locator

FAQ

Policy relieving the expert

4. Knowledge Services

deciding between relevant and irrelevant information due to many automated cognitive processes. This automation might also be disadvantageous, though, if experts experience difficulties to adapt to new problem settings or to accept new and revolutionary ideas. Experts spend more time to analyze the situation in difficult problem settings, are different from novices in their selection of problem solving strategies and are more able to control their cognitive processes than novices. The differences between experts and novices have substantial consequences for the design of knowledge infrastructures. This is especially true for functions such as personalization, system-supported recommendations and collaboration. Novices not only require a different presentation of knowledge elements than experts. Novices searching knowledge infrastructures for information on whom they could ask for help might need support by intermediates — participants just one or two skill levels above their own, not experts. The effort would be higher for both, for experts to reflect their decisions and for novices to learn from them. Experts on the other hand might be best “teachers” for workers at the skill level proficient and possibly competent. The expert’s knowledge is best analyzed with the help of story-telling. The expert should report critical situations together with context in which they occurred, subjective assessments of the situations and actions taken. Summing up, tutorials, case-oriented learning, exchanging stories and peer-to-peer learning deserves much more attention than the single-minded focus on experts teaching and answering questions of the rest of the employees. Experts can be supported by specific search services that aim at easing access to company or industry expertise in a certain domain. These are special cases of yellow pages that list contact details and a description of the expertise of knowledge workers who are considered experts in their domain. Lists of frequently asked questions (FAQ) are a popular instrument to store questions that might be of interest to many participants together with answers, mostly given by experts. Examples are the manyfold public FAQ lists that can be found on the Web. Most experts complain about being too busy to be able to answer all questions timely or, what is more, to express what they know in an FAQ list. An organizational policy might shift the workload of expressing answers to (frequently) asked questions from experts as the sender of the answer to the questioner as the receiver of the answer. Thus, a knowledge worker asking an expert would be held responsible for contributing question and answer into a FAQ system. In addition to the relieve on the expert the questioning knowledge worker, being lower in the ladder from a novice to an expert, might also find it more straightforward to convey the meaning of the answer in a form that other non-expert knowledge workers of a similar status can connect to more easily.

4.3. Interaction Services

4.3 Interaction Services Understanding group work and design of supportive ICT tools has been researched for over two decades under the topic of computer-supported cooperative work (CSCW, Hasenkamp/Syring 1994). It is an interdisciplinary research field and was started as an effort by technologists to learn from members of other disciplines, e.g., economists, social psychologists, organizational theorists and educators (Grudin 1994, 19f). Technical support may focus on various different aspects of interaction in group work as exchange of knowledge in shared workspaces, provision of communication media, structuring cooperative work processes and guidance of decision processes. Figure 4-13 shows classes of interaction services ideally all offered by an EKI. Every communicative relationship at some point is initiated. Initiation services thus target the systematic support of the establishment of links between people. Networking services target the deepening, renewal and care-taking for these links. Communication services focus on the exchange of knowledge between people and offer various different channels to enable or enhance this. Finally, community services target the support of groups that are characterized by long-term, social relationships with many rather weak links between their members. Consequently, these services form a cycle of increasingly closer forms of interaction as outlined above. Not every relationship evolves through all of these stages. The dashed line between community and initiation indicates that every community may be renewed by developing links to new people throughout these different stages. initiation

community

networking

communication Figure 4-13. Overview of interaction services

In the following, communication as an essential concept for interaction will be explained as a foundation of this chapter (section 4.3.1). Then, initiation (section 4.3.2), networking (section 4.3.3), communication (section 4.3.4) as well as community services (section 4.3.5) are discussed in more detail.

285

286

4. Knowledge Services

4.3.1 Foundation This section firstly outlines frameworks that describe different aspects of human communication and how they interrelate. Secondly, speech act theory is explained as an approach applied in the area of MIS that is suited to structure communication.

Shannon/Weaver’s communication theory

Communication models. There are numerous different models and theories that structure and explain human communication. One of the earliest is Shannon/Weaver’s communication theory. In the following, it will be shortly outlined and then turned towards other approaches that contribute and detail further aspects such as different levels of communication and the relevance of context. The theory by Shannon/Weaver (1949) is referred frequently in the context of MIS as a basic structure of communication processes. It distinguishes between source which is an entity that engages in the communication process, encoder that expresses the information to be transferred in form of a code, message composed of this code, channel that transfers the message, noise that may interfere the message such as physical transmission errors, decoder that reconstructs the information based on the code and finally receiver who represents the addressee of the message. Figure 4-14 visualizes the relationships of these components. -------------------------------------

noise

encoder

message

channel

decoder

source

receiver

Figure 4-14. Components of the Shannon/Weaver model

Reichwald’s communication model

Based on this model, communication would be defined as the transmission of information between a dislocated source and receiver by means of a code over a physical medium. However, this focuses strongly on two technically coupled systems and thus falls short to describe human communication which is much more complex. One critique is that his model does not acknowledge the sense-making process that is important for human communication. It mainly assumes that a sender packages an idea into a message and transmits it to the receiver. However, meaning cannot be transmitted this way but has to be reconstructed by the receiver (section 1.1.1, 4ff for transmission of information and knowledge). Therefore, Reichwald’s model distinguishes three layers of communication processes between a sender and a receiver, which can be both humans or machines (Figure 4-15). These layers are the same as the ones used in section 1.1.1 to delimit knowledge from related terms such as information, data and symbols.20 Physical signs are transmitted on the syn-

4.3. Interaction Services

287

tactic level, which can be compared to the channel in the Shannon/Weaver model. Similarly, typical factors that may disturb the communication process on this level are physical noise or interferences. On a semantic level, meaning is transferred by means of these physical signs. This requires their interpretation by the receiver. Possible errors that may occur associated to this process thus are the assignment of a wrong meaning to single signs or to sets of signs, e.g., sentences or words. The pragmatic level is concerned with concrete action and describes the communication of the sender’s intend. Errors include wrong reactions to messages, e.g., an urgent message is not handled with the appropriate priority. Overall, communication in Reichwald’s model is not only the transmission of physical signals but a process conducted in order exchange information required for action. From a business perspective, it is a basic foundation for the coordination of the distributed activities of an organization. pragmatic level: intention sender

semantic level: interpretation

receiver

syntactic level: transmission of signs communication processes

Figure 4-15. Reichwald’s (1993, 2176) communication model

As an extension of Reichwald’s model, different aspects of a message can be distinguished, i.e. factual content, self-revelation, relationship and appeal (Schulz von Thun 1981). Factual content describes the actual facts contained in a message. From this perspective, the sentence: “The traffic light has changed to green!” of a person sitting on the passenger’s seat just contains information about current traffic situation. Every message also reveals something about the sender, e.g., her way of speaking or his emotional state. The example sentence may reveal that the co-driver is in a hurry. A message also always characterizes the relationship between sender and receiver. In case of the example, it may manifest that the codriver does not agree with the driver’s way of steering a car. Finally, a message appeals to someone to do something, e.g., to accelerate the car in order to reach the cross-way in time. The factual, self-revelation and relationship aspects may be directed towards the appeal. Hence, they may be 20

In case of Reichwald’s model, the sigmatic and syntactic levels are combined.

Four aspects of a message

288

Communication & context

Speech acts

4. Knowledge Services

functionalized and do not necessarily express something “objectively” but mainly are a means of goal achievement. A second point that is criticized at the Shannon/Weaver model and transmission models of communication in general is that human communication is not linear and static. Rather, it should be regarded as a social process that strongly depends on context. Merten (1999, 103ff) in this regard defines communication as a social system that is characterized by reflexivity and selectivity. He distinguishes three types of reflexivity: factual, social and temporal. Factual reflexivity refers to the fact that a message only can be interpreted if it is embedded in a relevant context. The communicators form meta-assertions in order to make a message understandable and assignable, e.g., in order to distinguish between information and opinion or news announcement and commentary. Reflexivity on a social level means that the communicators adapt their messages to the social context, e.g., the role and competencies of the communication partner. Finally, temporal reflexivity describes the circumstance that communication process directly and indirectly refers to itself through feed-forward structures, i.e. the creation of expectations, and feed-backward structures, i.e. reactions to past communication acts. All three forms of reflexivity make clear that context does not only influence the interpretation of a message but also is reshaped by the communication process. Selectivity means that the receiver of a message needs to decide on the share of implicit or explicit communication as well as the parts of a message that are relevant for him. These decisions again are governed by various inner and outer contextual factors. Hence, it can be concluded that the inner and outer context is relevant for the analysis of communication processes. Speech act theory and its application. Speech act theory was developed in the 1960s in order to model human language. It focuses on the pragmatic level of communication and views speech as strongly oriented towards action. The theory divides human communication into single speech acts which represent verbal units that can be a word, a sentence, a paragraph, or an entire document (Austin 1962, Searle 1969). It investigates the meaning that speech acts can carry and the types of actions related to them. A speech act can be divided into three parts: sentence content that specifies the issue of communication, category that classifies the intention of the speaker, and presentation that specifies how the verbal units are presented to the listeners, e.g., polite, aggressive, determined. The category classifies the intention of the speech act as assertive, i.e. the speaker makes a statement about something, e.g., “Vulcans are characterized by pointed ears”, directive, i.e. the speaker wants to make the listener do something, e.g., “Could you please hand me the pencil?”,

4.3. Interaction Services

289

commissive, i.e. the speaker makes a commitment about something by the speech act, e.g., “I will drive you home”, declarative, i.e. the speaker declares concordance between sentence content and reality, e.g., “I hereby declare you husband and wife”, and expressive, i.e. the speaker describes a current psychological state, e.g., “I’m not very comfortable with that idea”. The articulation of a verbal unit is referred to as locutional act, its intention as illocutional act and the effect of the communication on the behavior of the listener as perlocutional act. Winograd and Flores (1986) applied speech act theory to short conversations with the goal to support it with ICT. Structures of conversations are described in a conversation network by interlinked speech acts. Arrows represent speech acts and circles indicate the state of the conversation. Conversation networks thus are state transition diagrams. Conversation networks are modelled on the type level, i.e. the actual content of utterance is not prescribed and depends on the actual context. It can have one or more final states where no further acts are expected by the conversation partners. All acts are linguistic, i.e. they represent a communicative action. The conversation network does not specify what a person should do or what consequences a speech act may have. Figure 4-16 shows an example of a conversation between a customer and an employee at the cash desk of a supermarket.

5

9

B: you need to weigh the vegetables

1

A: I would like to pay

2

B: cash desk is closed

6

A: ok, I will go to another desk

B: that makes x Euro

4

A: may I pay anyway?

3

Conversation network

A: here is the money

B: good bye

7

A: I do not have enough money

8

Figure 4-16. Example for a conversation network

The customer (A) comes to the cash desk with the intention to pay for her shopping. The cashier’s (B) answer depends on the actual situation.

#

conservation state

#

start/end state

290

4. Knowledge Services

The customer could either tell the amount of money she needs to pay, that the cash desk is closed or that she has forgotten to weigh the vegetables and needs to return. If the cash desk is closed, the customer can either ask the cashier to allow her to pay anyway at this desk or she can decide to go to another desk. When the customer knows how much she has to pay, she can hand over the money or realize that she has not enough money. If the cashier has received the money, he thanks and says goodbye to the customer. Conversations in groups consist of multiple sub-conversations between their members and thus of multiple potentially interwoven conversation networks. Conversation networks can be used to structure and organize communication processes in groups, e.g., with the help of structured workflows or software functions like group calendars that implement a pre-defined conversation network. An example for a standard language that defines speech acts on a technical level is KQML (knowledge query and manipulation language). It is used to permit communication between software agents (section 1.5.1, 47ff).

4.3.2 Initiation

Tasks supported by social software

Examples of social software

The initiation of communication is typically supported by the software class of social software. It is a rather recent concept, a subset of computermediated communication, that covers software that is used to create and maintain social networks or virtual communities. Typically, this category of software allows for easy-to-use mechanisms to create and maintain online profiles (social identity), build relationships and reputation (social capital), stay aware of a network’s activities (social presence), comment on and recommend to others (social feedback), interact with others (social interaction), organize physical meetings (social planning) and share content (social spaces) on the Internet. Social software focuses on supporting individuals voluntarily entering networks or communities. Therefore, it supports informal gatherings rather than formal organizational groupings in teams or workgroups which are typically focused by Groupware, project management and collaboration software. Hence, it is characterized by an informal, self-directed nature of joining networks. It has the potential of building larger and more effective networks. Examples for software that can be used with this aim in mind are easyto-use content management systems, such as text, audio and video Blogs, Wikis, fora, real-time communication, e.g., instant messaging or chat, and software platforms for rich interactions between its members that build on the friend-of-a-friend metaphor, such as the FOAF project, MySpace.com, Tribe.Net, Meetup.com or, with a business connotation, LinkedIn or Xing.

4.3. Interaction Services

Currently, a lot of organizations adopt these technologies and attempt to profit from them. However, it remains to be seen whether and how the additional challenges in business or organizational settings, particularly with respect to power distribution, incentive systems, data privacy and knowledge risks concerns, can be overcome. Figure 4-17 presents a screenshot of Xing21 as an example for businessoriented social networking systems.

Figure 4-17. Xing as an example for a social networking system

The stored and managed information includes contact information, personal interests and the curriculum vitae of participants. The system is able to visualize alternative routes between the current user and the participant currently displayed and offers means to browse links of the respective person to other people. Furthermore, it includes various interaction-oriented services such as an asynchronous messaging system, discussion fora and notifications, e.g., about changes within one’s individual network such as new contacts of people on one’s contact list.

21

URL: http://www.xing.com, last access: 2008-11-01

291

292

4. Knowledge Services

4.3.3 Networking

Address book

Above it was stated that networking services target the deepening, renewal and care-taking of links between people. These are fundamentally social activities. Technology can support those activities by establishing an infrastructure for the exchange and maintenance of related information such as appointments, contacts and information that supports keeping track of other’s activities, i.e. awareness information. One of the basic foundations for maintaining personal networks is handling contact information such as phone numbers, email addresses and street addresses. Essentially, it describes selected aspects of the person dimension (section 3.2.4). This is usually supported by an address book service. It can be distinguished between shared address books that contain, e.g., contact information about every employee of an organization or of external contacts relevant for an organizational unit or the whole enterprise, and individual address books that store collections of contacts for one person. Address book services also include functionality for the exchange of contact information, e.g., by means of directory services (section 2.3.1) or simply as an email attachments. This requires standardization of the set of attributes to be included as part of the contact information. vCard defined by the Internet Mail Consortium is a broadly-adopted standard that specifies the content of electronic business card files (http://www.imc.org/). Listing 4-2 shows an exemplary business card structured based on the vCard standard, which includes the person’s name, street address, phone numbers, affiliation and the date of revision. BEGIN:VCARD VERSION:3.0 CLASS:PUBLIC REV:2008-04-07 09:17:10 FN:John Miller N:Miller;John ORG:Example Org ADR;TYPE=work:Pine Str. 10;Springfield;1020 EMAIL;TYPE=internet,pref:[email protected] TEL;TYPE=work,voice:1-800-416-55343 TEL;TYPE=cell,voice:1-900-416-33232 END:VCARD Listing 4-2.

Calendar

Example of a business card based on the vCard standard

Whereas address books concentrate on the person dimension, calendars focus on the time dimension and support the management of individual or shared appointments. A basic calendar item is specified by subject, start date, end date, location and description. In addition to that, such an item

4.3. Interaction Services

can serve as invitation by adding required or optional participants and sending them a meeting invitation that can be accepted, denied or answered with an alternative proposal. In order to find a suitable time slot for the appointment, a free/busy lookup can be made to the personal calendars of the invited people. The standard for exchange of calendar entries and appointments is iCalendar22 or short iCal, a text-based format that distinguishes events, free & busy time, journal entries and to-do items (see section 3.2.4). Electronic calenders need to support the definition of access privileges, since they usually contain sensitive information. This comprises at least determining who has access and which calendar items are not accessible at all. Group awareness refers to the perception of what other members of a group currently do or have done in the past. It is defined as an understanding of the activities of others which provides a context for own activities (Dourish/Belotti 1992). One of the first notions of awareness was in a study that observed work in a London Subway control room. It researched the question of how people in a group know when to do what. It emphasized the importance of implicit cues about the activities and the role of subtle communication acts like mimics and gestures which was coined “periphal awareness”. Awareness functions are seen as a way to make coordination more efficient as users of EKI often are spread over several distant locations. Different modes of awareness can be distinguished based on whether it is focused on past, present or future events in a system and whether these events are tightly or loosely coupled to the actual work context. Table 4-4 shows exemplary questions for each of the resulting modes of awareness. For tight coupling, usually more detailed information is required. Awareness information should be provided as unobtrusively as possible which is also referred to as “peripheralness” of the way of presentation. Table 4-4.

293

Group awareness

Modes of awareness

Group awareness modes past

present

future

tightly coupled

What changes happened on the draft that I uploaded last week?

Which member of my project team is currently online?

Who will read my project report?

loosely coupled

How many documents were added last month?

What are the activities of other participants in the system?

How will other workspaces evolve?

Positive results of group awareness are that it encourages spontaneous, informal communication as well as creation of a shared culture since 22

iCalendar is specified by the IETF RFC 2445.

294

Example BSCW

4. Knowledge Services

group members are better informed about activities, state and workload of other group members. Moreover, all participants stay informed about the current state of collective team activities. Therefore, members are better able to judge whether individual contributions are relevant to the group activity and evaluate their own actions with respect to the group’s goals and progress. However, awareness information may clutter the user interface with a lot of information not directly necessary or useful. Hence, it must be decided thoroughly in which case users need what type of information. Awareness functions also rely heavily on monitoring and capturing of data regarding to user behavior, which may contradict data privacy. Therefore, it should always be transparent to users what information is recorded. Figure 4-18 shows a screenshot of OrbiTeam Basic Support for Cooperative Work (BSCW), a Web-based collaboration system. Icons visualizing user events are assigned to items within the system such as a folder or a document. If a user clicks on one of these, she gets more detailed information, e.g., a list of users who recently accessed or read the item.

Figure 4-18. Awareness within OrbiTeam BSCW

4.3.4 Communication Communication services have different characteristics and thus they are not similarly suited for all kinds of communication tasks. When designing an EKI it is relevant to estimate when users would perceive which type of

4.3. Interaction Services

communication service as most appropriate. Two theories that try to explain related decisions are media richness and media synchronity. These will be outlined first as foundation before turing to the actual basic communication services. Media richness theory, also called information richness theory, was formulated by Daft and Lengel (1986) and poses that media vary in terms of their ability to enable users to communicate and change understanding, i.e. they are different with regard to their “richness”. A media is richer if it has a greater language variety, i.e. the ability to communicate natural language instead of numeric information, a greater multiplicity of cues, i.e. the amount of ways how information can be transferred such as written text or voice, a greater personalization, i.e. the ability to individualize a message, and a better support of rapid feedback. In these terms, direct physical interaction is much richer than interaction by means of traditional mail. It decreases from video conferences over phone communication to email. It is proposed that task performance is higher when the media’s ability to convey information is matched with the task’s needs. Certain media are better able to transmit information depending on whether the information is used in situations of uncertainty or equivocality. Uncertainty results from a lack of information but nevertheless there is a framework available for the interpretation of a message. Equivocality exists when there are multiple interpretations for a message or frameworks available that the receiver of a message may choose from. Daft and Lengel argue that rich media are better suited for equivocal tasks whereas “leaner” media are best suited for tasks of uncertainty. However, critics state that media richness does not take into account other factors that determine media choice, particularly social norms, user experience, a critical mass of communication partners and the general organizational context. The theory was not supported by the results of several empirical studies (e.g., Dennis/Kinney 1998, 269f). As a consequence, the theory of media synchronicity has been developed (Dennis/Valacich 1999). It is based on the assumption that the attainment of work goals is determined by two processes: the transfer of contents, labelled with conveyance, and the attainment of a consensus, referred to as convergence. Hence, media synchronity theory does not focus on the general task that should be performed but rather on microlevel communication processes that need to be accomplished by every individual. Media characteristics may influence these processes positively or negatively. The relevant characteristics identified by Dennis and Valacich are immediacy of feedback, i.e. the extent to which a medium enables users to give rapid feedback on the communications they receive,

295

Media richness

Media synchronicity

Characteristics

296

4. Knowledge Services

symbol variety, i.e. the number of ways in which information can be communicated which subsumes multiplicity of cues and language variety of media richness theory, parallelism, i.e. the number of simultaneous conversations that can exist effectively, irrepressibility, i.e. the extent to which the media enables the sender to edit or fine tune the message before sending, and reprocessability, i.e. the degree to which a message can be reexamined or repeatedly processed within the context of a communication event. For example, media that allow for a high immediacy of feedback but only a low parallelism are supposed to be well-suited for convergence processes. Table 4-5 categorizes selected media with the help of these characteristics. For example, while an electronic email has the advantage that it can be fine-tuned by its editor and easily reproduced by its receivers, his communication partners cannot respond as quickly to it as within a video conference - that in turn only allows for low parallelism and reprocessability. Dennis and Valacich conclude that it depends on the given situation whether a medium is the “best” one. Hence, a general classification of richness as proposed by media richness theory is not worthwhile. Both, conveyance and convergence processes can be observed simultaneously in relation to a specific task and thus it might be most appropriate to switch between different media. Table 4-5.

Characterization of media according to Dennis/Valacich (1999)

face-to-face

immediacy of feedback

symbol variety

parallelism

irrepressibility

reprocessability

high

low-high

low

low

low

video conference

medium-high low-high

low

low

low

telephone

medium

low

low

low

low

written mail

low

low-medium

high

high

high

voice mail

low

low

low

low-medium

high

electronic mail

low-medium

low-high

medium

high

high

online chat

medium

low-medium

medium

low-medium

low-medium

asynchr. Groupware

low

low-high

high

high

high

synchr. Groupware

low-medium

low-high

high

high

high

Classification of communication

Communication services are classified based on the following criteria: time, i.e. whether messages are exchanged at the same time or at different times, which is also referred to as synchronous vs. asynchronous communication,

4.3. Interaction Services

297

place, i.e. whether the communication partners are located at one or at different places, number of communication partners, i.e. whether two or more people participate in the communication process, which is sometimes also described by the classes of one-to-one, one-to-many and many-to-many communication. Further criteria are predictability of time and place, size of the group of communication partners, control models as well as coordination processes used to structure group work (Bafoutsou/Mentzas 2002, 282ff). However, the ones enlisted above are used frequently in order to classify services of technical infrastructures. Consequently, the services outlined in the remainder of this section will be classified based on the distinction between services for asynchronous and synchronous communication. Synchronous communication services. Synchronous communication is supported by a variety of media such as chat, instant messaging, phone, teleconferencing systems and video conferencing systems. At this point, we give a short overview of the most important communication services offered by an EKI. Chat enables two or more people to conduct text-based, synchronous conversations. They take place in so-called chat rooms or channels. These channels may be dedicated to selected topics. In many cases, users are allowed to log-on anonymously using a nickname that they are free to choose. A chat can be limited to rather basic text-based channels but may also offer many further functions, e.g., for one-to-one secure communication, file transfer and user directory services. Today, many chat systems are implemented based on Web-based interfaces and thus are accessed by means of Web browsers. However, chat services may also be offered by specialized servers where users connect to by means of a client software. Internet relay chat (IRC) is an example for a protocol that is used by many public chat networks like Efnet, Undernet, IRCnet, DALnet or NewNet.23 However, instant messaging for one-to-one conversations became more and more popular during the last years. Nevertheless, an advantage of chat is that it also enables many-to-many communication within a group of people. Instant messaging (IM) is focused on one-to-one communication via text-based messages and is suited well for short, informal communication. Users are required to install an (often freely available) IM client on their local computer and need to create an IM account on one of the servers operated by an IM provider. Many IM-networks are implemented based on

23

The original IRC standard is described in the Request for Comments (RFC) 1459 followed by four newer RFCs 2810-2813, URL: http://www.irchelp.org/irchelp/rfc/, last access: 2008-10-30

Chat

Instant messaging

298

(IP-)Phone

Audio conferencing

Video conferencing

4. Knowledge Services

an assisted peer-to-peer architecture: centralized servers support lookup of peers and their current status. The actual communication data is exchanged directly between the peers. An advantage of IM are awareness functions that inform users about, e.g., the online/offline status of communication partners. These are enlisted on a so-called buddy list. This enables ad-hoc, informal communication. There are multiple competing IM standards, clients and IM networks from different companies like AOL, Mirabilis (ICQ), Microsoft and Yahoo. However, there are also efforts for the creation of common standards in this area. The Extensible Messaging and Presence Protocol (XMPP) is an XML-based standard that acts as a foundation for the creation of interoperable protocols for instant messaging, multi-user chat or the exchange of certificates.24 Phones are still one of the most important communication media for one-to-one voice-based communication. Video phones today have no considerable market share, though their technical implementation has been solved years ago and they have been offered on the market for quite some time. Recently, IP telephony has gained increasing attention in many organizations. It denotes a set of facilities for synchronous exchange of voice, fax and other forms of communication using the Internet Protocol (IP, section 2.1.3, 89ff) that have traditionally been carried over circuit-switched connections of commercial or government-owned telephone networks (PSTN). Advantages of IP telephony are that it avoids the tolls charged by private or public owned telephone companies and that it can be run on the same network infrastructure as other network technologies. Disadvantages are that IP telephony over public packet-switched networks may have a low quality as it is not ensured that data packets arrive in an acceptable time-frame. A better quality of service is possible with private networks managed by an enterprise or by a service provider. Audio conferencing systems enable multiple individuals to join a voicebased conference. This may be implemented as a conferencing system where all participants connect to by dialing a phone number. Alternatively, one person step-by-step phones up the other participants and connects them to the conference. This way of communication typically offers no visual cues that help to coordinate the individual contributions, e.g., gestures or facial expressions. Hence, the teleconference needs to be conducted in a disciplined way. Specialized conferencing services may be used to support the coordination of teleconferences, e.g., by issuing the right to speak to participants that afore “raised a hand”. Video conferencing services enable interactive, bidirectional communication based on voice and video. Compared to text-based or voice-communication, they offer a “richer” communication channel as visual impres-

24

XMPP is defined by several RFCs, e.g., RFC 3920 to 3923. See http://xmpp.org/rfcs/, last access: 2008-10-30 for further details.

4.3. Interaction Services

sions of the communication partners can be formed and nonverbal communication is enabled. However, in everyday business life audio conferences are often preferred. Professional video conferencing systems transmit voice and video with high quality, but usually cause high costs as separate systems need to be installed. Desktop video conferencing is a less costly alternative. It is based on personal computers that are prepared with a Web cam, a microphone and a software client. Voice and video quality are substantially lower due to limited network bandwidths of public networks. One of the dangers therefore is that facial expressions cannot be easily recognized. Standards that define network communication protocols for video conferencing are H.320 for conferences over ISDN and H.323 for IP-based networks. Examples for software products are Adobe Acrobat Connect25 that is based on Web-based technologies and the Skype client26. Driven by IP-telephony, synchronous communication means are converging today. A number of software vendors offer so-called unified messaging servers, that support instant messaging, IP-telephony and desktop video conferencing combined in a single client application and integrated with an enterprise user directory. IBM Lotus Sametime and Microsoft Office Communication Server are examples that also add many-to-many communication possibilities and provide presence information. Application sharing services enable two or more participants to access one and the same application by means of a joint interface. In most cases, the application is installed on one of the participant’s computers such as a text processing or spreadsheet software. Typical situations where application sharing services are used are the joint review of text documents by the authors, the training of the usage of a software or the troubleshooting by a help desk employee on a remote computer. Principally, application sharing also includes software that is specialized on selected collaborative tasks, e.g., electronic whiteboards for the creation of joint drafts during electronic meetings or co-browsing software for joint Web access. Remote desktop software like VNC27 can also be summarized under this headline. Meeting support systems combine multiple other services such as address books, calendars and audio or video conferencing. They help to structure and support tasks related to organizational meetings, e.g., the invitation to appointments and the distribution of messages associated to a meeting. They may also include formalization services since meetings are often associated with two types of documents: agendas that structure the topics discussed during meetings and protocols that document their results, e.g., decisions taken and follow-up actions to be accomplished. An example for such a service is WebEx28. 25 26 27

URL: http://www.adobe.com/products/acrobatconnect, last access: 2008-10-30 URL: http://www.skype.com, last access: 2008-10-30 URL: http://www.realvnc.com, last access: 2008-10-31

299

Unified messaging

Application sharing

Meeting support systems

300

Email

List server

4. Knowledge Services

Asynchronous communication services. EKI can include a rich variety of systems for asynchronous communication such as email, list servers and newsgroups. Email is one of the most wide-spread used applications of the Internet and supports one-to-one as well as one-to-many communication. Today, email client programs are usually based on POP3, SMTP and IMAP (section 2.2.3). Emails are composed and received by means of email clients, which are also called mail user agents (MUA), that are installed on local PCs or by means of Web interfaces. Examples for local programs are Mozilla Thunderbird, Apple Mail or the mail components of Groupware clients such as Microsoft Outlook or the IBM Lotus Notes client. Webbased email interfaces offer comparable functionality including an email inbox, folders for the storage of ones emails, filtering rules for automatic categorization of messages, spam protection filters and so on, especially when using AJAX (see section 5.2.1, 374ff). Many organizational email infrastructures can be accessed either by an email client or by Web interfaces in order to enable uses to access their emails from external locations and allow for seamless integration of the mail inbox into the employee portal. Today, email substitutes many paper-based mails and also many phone calls. Compared to the traditional mail system, email is much cheaper, faster with respect to delivery times and creation of messages, less formal and more straightforward. However, it also implies new challenges such as spam or junk mail, i.e. undesirable mass postings of emails, an increasingly large amount of messages that has to be handled by single users as well as the threat of spreading of viruses, worms and trojans via email. List servers, also called email distribution lists, support one-to-many broadcasting of messages by managing mail subscription lists and distributing emails to the addresses on these lists. The distribution list can be managed via email, e.g., by sending an email with the instructions “subscribe” or “unsubscribe” in the subject line or email body to the emailaddress of the respective list in order to join or leave it. Application fields are the distribution of newsletters and communication within communities. The ISWorld mailing list is an example of an email service that connects several thousand MIS researchers all over the world and has become an important medium to announce conferences, distribute calls for research papers, pose requests to the community or generally discuss topics with colleagues about research and teaching in the field of IS.29 Listserv, mailman and majordomo are examples for software programs that implement listservers.

28 29

URL: http://www.webex.com, last access: 2008-10-31 URL: http://lyris.isworld.org/isworldlist.htm, last access: 2008-10-30

4.3. Interaction Services

Newsgroups, also called blackboards or discussions, are forums that allow for many-to-many text-based communication about selected topics. Messages are structured in threads, i.e. hierarchical trees of messages and their replies. Discussions are an important technology to support communication within communities. Traditionally, the Network News Transfer Protocol (NNTP) has been used for the communication in newsgroups which requires a newsreader client. A large network of public newsgroups is represented by the Usenet. It is hierarchically structured according to selected discussion topics. The top level consists of the categories such as biz for business, comp for computer, misc for miscellaneous and talk for current issues and debates. Today, many alternatives exist for the realization of newsgroups which often offer Web-based interfaces. Some Internet portals or search engines offer services for the creation of groups and also the keyword-based search or browsing through groups including the Usenet, e.g., Google Groups30 and Yahoo Groups31. Overall, there is a rich diversity of Internet-based services that can be used for the exchange of messages by means of newsgroups.

301 Newsgroups

4.3.5 Community Community services can be distinguished into services that focus on the creation and exchange of documented knowledge and systems that should provide a general platform for the support of work in groups. The former will be described in the following under the topic of collaborative authoring before turning to Groupware platforms. Collaborative authoring. Joint or collaborative authoring is the synchronous and asynchronous creation and exchange of principally any type of contents, but normally focused on text-based documents. It can be supported by services such as application sharing, functions offered by current text processing software for the addition of comments and monitoring of changes applied by other users as well as by more advanced services such as multi-user editors and Wikis, which are described in the following. Multi-user editors facilitate a joint, synchronous writing process within a shared editor in contrast to an asynchronous writing process where a document is extended and changed individually and subsequently is exchanged with co-authors by means of, e.g., email or shared information spaces. This requires enhanced coordination functions, e.g., for the locking and change of complete paragraphs or even of single sentences, for the highlighting and approval of changes by co-authors or for the notification of relevant events and activities. Usually, the co-authoring process is backed by additional communication services that allow for synchronous 30 31

URL: http://groups.google.com, last access: 2008-10-30 URL: http://groups.yahoo.com/, last access 2008-10-30

Multi-user editor

302

Wiki

Groupware

4. Knowledge Services

communication to share opinions, reach consensus and coordinate activities. However, though there are some research prototypes, multi-user editors so far are not wide-spread used in practice or subsidized by text processing software that is shared by means of application sharing services. Product examples are MoonEdit, SynchroEdit and Google MobWrite. Wikis are Web-based systems targeted at an easy and quick addition of own contributions to the overall content-base. Therefore, a Wiki can be defined as a system that offers functions for the management of a freely expandable collection of modifiable interlinked Web pages (Leuf/Cunningham 2001). Consequently, Wikis are a special class of Web content management systems (WCMS, section 4.4.4, 321ff). Typical for a Wiki is that principally every user is allowed to create and edit pages and thus is involved in an ongoing process of collaborative, asynchronous content creation. In the pure Wiki approach, there are no user privileges or access limitations. This is their main difference compared to traditional WCMSs that offer functions for the management of privileges and also structured publishing processes. Nevertheless, Wikis are increasingly used in an organizational context, e.g., in domains where contents published are not sanctioned by legal restrictions or pre-defined forms such as software documentation or the description of formally implicit organizational procedures in a bottom-up manner. A well-renowned example for a Wiki Web site is Wikipedia, a meanwhile large public encyclopedia created by Internet users that is available in different languages.32 Groupware. Groupware can be generally denoted as information systems relevant in the context of CSCW. The following part gives a more detailed description and also includes a comparison of Groupware and workflow management systems (WfMS). Furthermore, group decision support systems (GDSS) will be explained as a class of systems specialized at the structuring of decision processes as well as shared information spaces used as a general metaphor for applications that should enable groups to exchange information and knowledge. Groupware comprises all computer-based systems that support groups of people engaged in a common task (or goal) and that provide an interface to a shared environment (Ellis/Gibbs/Rein 1991, 40). It provides general support for collection, organization and sharing of information over network infrastructures within distributed collectives of people, e.g., work groups and project teams. Groupware focuses unstructured, flexible processes in workgroups. Typical functions are email support, group calendars, task lists, directory services and shared document collections. Therefore, Groupware combines many of the services that have been described above individually. Examples for Groupware servers are IBM Lotus Dom-

32

URL: http://www.wikipedia.com, last access: 2008-10-30

4.3. Interaction Services

ino33 and Microsoft Exchange34. It also supports information replication between servers and clients, which enables each participant working offline on local copies. When these users later on connect to the server, changes made to, e.g., calender items, contact information or forms, are synchronized with the server’s database and hence with the other clients. A fundamental feature of Groupware is the focus on ad-hoc, flexible processes. Therefore, focus on a different type of processes compared to traditional workflow management systems (WfMS) (section 3.4.2, 225ff) as they focus on recurring and structured processes and tasks. Whereas Groupware targets the flexible support of group tasks and processes, WfMS are about the efficient coordination of complex yet formal processes. The main objects relevant for Groupware are individuals and groups. WfMS in contrast focus on the coordination and control of tasks and workflows. Thus, it represents the source of activity and control whereas connected to Groupware, this in most cases is the user and not the system. Groupware systems due to their bottom-up nature can be built-up in a decentralized manner, e.g., by a number of synchronized servers. WfMS in contrast can be characterized as mainly centralized systems. Nevertheless, it should be noted that Groupware platforms often also incorporate workflow functionality in order to support structured tasks such as the revision of documents. These workflows are usually ad-hoc and less complex than in traditional workflow systems. A group decision support system (GDSS), also called group support system (GSS) or (electronic) meeting support system, is an interactive system that combines communication, computer and decision technologies to support formulation and solution of unstructured problems in group meetings. Participants may be geographically distributed but usually are located at the same place such as a meeting room that is extended with hardware like electronic whiteboards and computer screens for each individual participant. GSS integrate technologies to support communication in groups, structuring of processes by which groups interact, e.g., agenda setting or facilitation, and information processing, e.g., information aggregation, evaluation and structuring. GSS can be classified according to the level of support in level 1 GSS which remove communication barriers, level 2 GSS that provide decision modeling and group decision techniques and level 3 GSS that provide expert advice in selecting and arranging rules in a meeting and thus lead to machine-induced group communication patterns. An example for a GSS is GroupSystems (DeSanctis/Gallupe 1987, Zigurs/Buckland 1998). Shared information spaces are virtual places where common resources such as documents or links can be stored and manipulated. They also may

33 34

URL: http://www.lotus.com, last access: 2008-10-30 URL: http://www.microsoft.com/exchange, last access: 2008-10-30

303

Groupware vs. workflow management system

Group decision support system

Shared information space

304

4. Knowledge Services

offer functions for interaction and collaborative management of documented knowledge. Ideally, shared information spaces are workspaces that integrate all functions to store, retrieve and jointly create information resources as well as to coordinate activities and manage group processes within geographically dispersed groups. A shared information space can be regarded as a metaphor that is used to combine a set of Groupware functions for sharing and communicating documented knowledge and thus principally may refer to many different types of concrete applications or systems. These shared information spaces can be organized on a central collaboration server (e.g., Microsoft Sharepoint Server, or ICEcore) as well as with decentral P2P systems (e.g., MS Office Groove). Example: Microsoft Office Groove. Since the acquisition of Groove Networks by Microsoft, the product Groove is sold as part of Microsoft’s Office suite. It targets the support of interaction in small groups with two to about 30 members. Collaboration is organized within workspaces that can be accessed with a Windows client. These may contain a variety of tools for managing contents such as text notes, documents, calendar items or images and for collaboration such as group calendar, contact list, discussion forum, meeting minutes, task list and sketchpad. Groove emphasizes the support of ad-hoc collaboration in groups, e.g., by offering a buddy list that shows the online status of other workspace members, cobrowsing functions for synchronized navigation through the Web, a navigate together option that synchronizes the interface of the group workspace and awareness services that visualize information about current activities of other users, e.g., the tools they access at the moment. Besides the files tool, which offers basic functions to store files and to categorize them into hierarchically structured folders, Groove offers a Sharepoint files tool that connects a Groove workspace to a Sharepoint document library and thus to a centralized file store that includes document management functions such as versioning. Groove is not only interesting as it includes a number of collaborative tools but also because it is based on decentralized peer-to-peer architecture (see ”Peer-to-peer architecture” on page 51). Groove peers communicate directly with each other and do not require a central server. The workspaces are stored on each peer in an XML-based format and synchronize automatically as soon as the peers connect via the network. Device presence services handle the detection of other peers and their online status. Peer connection services determine IP addresses of other peers and handle communication using the proprietary simple symmetrical transmission protocol (SSTP). Nevertheless, centralized servers can be deployed optionally for administrative tasks such as directory services, distribution of software updates, monitoring storage space and network usage and to ensure the synchronization of peers with relay services, e.g., if they are located behind a firewall.

4.4. In-form-ation Services

4.4 In-form-ation Services This chapter describes EKI services that support the creation of documented knowledge, i.e. bringing knowledge into a form so that it can be easily distributed and reused. This process can be labelled as “formalization” since knowledge is expressed by means of, e.g., written language, graphical representations and formulas, which involves a more or less strong formalization process. This type of services is referred to as “inform-ation” services in order to emphasize that individual knowledge is transformed to contextualized information instead of knowledge as soon as it is expressed and incorporated by some type of information resource (section 1.1.1). Figure 4-19 displays four general groups of in-form-ation services which are discussed in detail in this chapter. Capturing refers to the process of making potentially large amounts of information resources electronically available in the format and quality as required by tasks in organizational business processes. This comprises scanning of paper-based information resources and further processes such as visual post-processing and optical character recognition (OCR). It may also include the conversion and loading of legacy data that already exists in various electronic formats. Manipulation services support the coordination and management of changes on information resources. Translation services concern the transformation of information resources into different structures and formats. They serve the important task to publish information on the Intranet or on the Web. Archiving services help managing different retention periods of information resources, storing them in a secure and cost-effective manner and ultimately assist their deletion. They are relevant because legal regulations oblige organizations to keep information resources for defined amounts of time and to be able to provide them as an object for investigation and piece of evidence. capturing

archiving

manipulation

translation Figure 4-19. Overview of in-form-ation services

The order of these groups reflects a life-cycle of information resources from their creation to the point of their deletion. The loop visualized with

305

306

4. Knowledge Services

the dashed line in Figure 4-19 indicates that information resources might be re-used for reference and creation of new information-resources. The following section deals with the definition of the most important terms and types of systems as a foundation for this chapter (section 4.4.1). Afterwards, capturing services (section 4.4.2), manipulation services (section 4.4.3), translation services (section 4.4.4) and archiving services (section 4.4.5) are presented.

4.4.1 Foundation

Information resource life-cycle

Relevance of stages

In-form-ation services deal with the creation and management of resources that represent explicit knowledge. These resources have a life-cycle which will be outlined firstly as a general foundation of this chapter: (1) Creation is the generation of information resources. This may involve other information resources of various types that, e.g., act as templates, are re-used (partly) as building blocks and are applied for reference or for information. (2) Revision is a life-cycle stage where information resources are subject of extension, change, annotation and correction. The revision and extension process may involve one or many authors so that interaction services may also become important here. (3) Stability denotes the circumstance that information resources are not subject to frequent changes. A formal approval process might be conducted in order to transfer contents into this stage where their quality and adherence to formal criteria is ensured. If changes are required later on after approval, they are incorporated by new and distinguishable releases. (4) When information resources are not actively used anymore but need to be retained for a defined amount of time, e.g., due to legal regulations, they enter into the archived stage. During this stage, they are regularly moved to and retained on separate storage media that are cheaper yet slower than storage media used. In contrast to the stage of stability, documents are accessed only rarely. (5) Lastly, information resources are deleted when they are not required anymore or when legal regulations permit deletion. In practice actually, for certain types of information resources it may not be the primary question how long documents can be retained but at which point of time they can be deleted earliest as they might incorporate legal obligations. The amount of time that an information resource remains within one of these stages can be differently depending on factors such as its type. A specific stage might also be irrelevant at all. The transition between two stages can be tied to a specific event or point in time, e.g., a document turns into stability stage when a review and acknowledgement process has been completed and afterwards the organization is allowed to delete it after keeping it for twelve years. Nevertheless, the boundaries between two stages can be blurry, e.g., for co-authored documents the transition between the creation and revision stages can only be set very roughly, e.g.,

4.4. In-form-ation Services

to the point in time where an initial and individually created version of a document is distributed to co-authors, who then start with extensions. In the introduction to this chapter above, the sequence of in-form-ation services is based on the life-cycle stages of information resources. Capturing services deal with the import of information resources and thus correspond to the stage of creation. Annotation services are used for complementing information resources with meta-data and thus are predominantly relevant for the stages of creation and also revision. Translation services deal with transforming information resources into different types of formats. They are particularly relevant during the stage of stability because agreed upon information should be published over different media and in different formats. Archiving services deal with the efficient retention and timely destruction of information resources and thus correspond to the stages archived and deletion. Principally, services from all different groups can be accessed during each stage of the information resource lifecycle. The correspondence should be seen only on a general level.

307

Relation to inform-ation services

Information resource-related terms. A life-cycle as described before can be identified for many different types and formats of information resources, e.g., university bachelor thesis, meeting protocols, Web pages, invoices and also for Wiki articles and emails. In theory and practice, the terms document and content are often used synonymously as general terms in order to refer to different classes and types of information resources. However, it is worthwhile to clearly separate the different terms as they emphasize different aspects in this area. An information resource is anything that can be addressed and manipulated by a human or a system as a discrete entity. Information resource is the most general term for any resource that can be managed with EKI services. In terms of the W3C, a resource is anything that might be identified by a URI and that can be conveyed in a message (section 3.1.2).35 It can be stored on non-electronic media such as paper or microfiche as well as in electronic formats. From the perspective of EKI, predominantly semi-structured information resources are relevant as they are suited to represent explicit knowledge, though the term principally may include both, strongly structured ones that follow a clearly defined data definition schema such as records in a database as well as semi-structured resources. A specific information resource can be composed of multiple other information resources, e.g., a spreadsheet document may contain tables, formulas, diagrams and images. The term information resource thus may be used to denote atomic entities on a fine-

35

URL: http://www.w3.org/TR/webarch/, last access: 2008-10-30

Definition of information resource

308

4. Knowledge Services

grained level as well as complex objects composed of multiple other information resources, bound by a specific purpose and accessible as a whole. The term resource emphasizes that the information in question is useful to be applied in order to achieve an outcome desired. Examples for information resources are emails, text processing documents, spreadsheets, scanned images, audio and video files, all in various formats such as plain text, OpenDocument format and PDF in case of text processing documents. Characteristic of basic resources such as energy, land and water is that they are scarce. This seems not to be valid here when considering the vast amounts of information that today are easily available for individuals and organizations in western societies. However, relevant is only the right information in the right quality. Furthermore, it is also the information processing capacity of individuals and organizations that limits the amount of acquirable information so that is still correct to speak of the right, appropriate and processed information as a scarce resource. Definition of content

Content is a set of meaningfully arranged electronic data that is separated from technical structure and layout. The term content separates the actual subject matter described by some information resource from technical structure and layout. Technical structure refers to a certain file format or schema used to store content. Nevertheless, content still has a semantic structure that, e.g., may describe the relationships between single components such as the reading order of headlines and paragraphs of a text document. Layout describes the rendering of content during reproduction. For example, one and the same content such as a software documentation can be rendered as HTML file in order to be displayed in a Web browser, as WAP file for mobile appliances, as PDF file to be printed out and in a format used as input for braille keyboards. Content itself cannot be displayed, changed or stored directly as it is always represented by means of information resources. Thus, it is an abstract concept often used when it should be emphasized that one and the same content can be distributed over different media, e.g., in the context of cross-media publishing or Web content management (section 4.4.4).

Definition of asset

Asset denotes a subset of content that is annotated with meta-data in order to enable its retrieval and (re-)use. The term asset emphasizes the value that a specific amount of content represents for an organization. In order to be valuable, content needs to be easily retrievable and reusable. This can be achieved by complementing contents with meta-data that describes relevant aspects, which is supported by annotation services. For example, if meta-data describes the authors, software modules affected and target groups of a software documentation,

4.4. In-form-ation Services

309

it can be automatically referenced on Web pages about specific software modules or can be retrieved by searching for the author’s name. A document is a legally sanctioned record of a business transaction or decision. The term record emphasizes that a document retains some kind of business transaction or decision. Legal regulations pose requirements that have to be fulfilled by documents concerning format, period of retention and way of storage. Typically, documents need to be stored in a format that is independent from specific applications, need to be conserved over several years in a way so that it can be proofed that they were not altered somehow. Typical examples for documents are purchase orders, contracts, invoices, patents and certificates. However, depending on the relevant regulations and the industry of an organization, the number of legally sanctioned records can be very large. For example, organisations in the pharmaceutical industry are obliged to archive every document relevant in relation the development of new drugs, which also includes emails and thus a substantial amount of data to be archived. The term document is frequently used in a much broader definition that is closer to information resources as defined above. For example, files created with common office applications such as Microsoft Word or OpenOffice Writer are commonly denoted as documents. Similarly, the term content sometimes is used to denote all information resources of an organization. Content is also used to denote that not only office documents and text has to be managed, but also multimedia resources like images, audio- and video resources. This implies the danger that the terms information resource, content and document are used largely synonymous, which fails to explain the main foci and exact meaning of these concepts.

Definition of document

Broad interpretation of the term

Management of information resources. Just as there are different and often vaguely-defined denotations for information resources, multiple different terms are used for their management in an organizational context as well as for the systems applied in this context. Therefore, the document management (DM) approach will be defined first and then the differences and similarities to other approaches will be highlighted. Document management is the systematic, goal-oriented and enterprisewide formulation, evaluation and enhancement of document-related processes over the entire information resource life-cycle, i.e. capturing, annotation, change, storage, review, archival and deletion of documents. The definition of DM focuses on documents as defined above in a narrow sense, i.e. legally sanctioned records that reflect a business transaction

Definition of document management

310

Content management

Enterprise content management

Records management

Definition of document management system

4. Knowledge Services

or decision. The definition deliberately places its focus on the documentrelated processes to be supported, which usually are part of other business processes of the organization. Hence, DM is not only concerned with the introduction of technical systems but focuses on the management of a specific aspect of the overall organizational business activities, i.e. the systematic and efficient handling of documents in order to guarantee traceability of business transactions. The term content management is also used in this context but a clear delimitation from DM is not possible as it is often applied based on a broad understanding of the term document closer to information resources. Web content management is a specialized discipline that concentrates on the translation and publishing of Web contents and will be discussed in detail in section 4.4.4. The terms enterprise content management (ECM) and enterprise DM are used to emphasize the need for an enterprise-wide approach as required by the definition above. It reflects the circumstance that many regulations and systems in fact only serve needs of single departments or functions but fail to support organizational business processes in an integrated manner. Though the integration of all document-related systems might not always be necessary, ECM represents a vision that at least should guide all local DM initiatives. The term records management (RM) is often used to emphasize the need for compliance (section 4.4.5), for managing retention and deletion of documents at the end of their life-cycle. RM has also been used to refer to systematic documentation of business transactions. Essentially, it can be used and will be used in this book synonymously to the narrow definition of DM described above. It should be kept in mind that DM is often used more broadly for the systematic management of information resources, especially in Anglo-American countries. Requirements and guidelines for RM are formulated in the standard ISO 15489, which describes standards for record keeping in public and private organizations as well as more general requirements for transparent and traceable administration and business processes. Within the recent discussion, deletion of documents is typically added as a task that has to be systematically managed, presumably due to the increasingly larger amounts of documents that have to be recorded and archived due to legal regulations such as the Sarbanes-Oxley act that demand a more comprehensive traceability of organizational activities and thus includes new types of electronic documents such as emails. Document management systems (DMS) provide services to support all processes related to the management of semi-structured information resources over their entire life-cycle, i.e. their capturing, annotation, change, storage, review, archival and deletion.

4.4. In-form-ation Services

311

This definition is closely based on that of DM as DMS represent the system class typically applied in the context of DM initiatives. However, it is not only focused on documents in the sense of legally-regulated entities but generally refers to semi-structured information resources. The reason is that the management of semi-structured information resources requires similar technical services such as versioning, check-in / check-out mechanisms, use of access privileges and annotation with meta-data, put aside specific functionality for ensuring integrity and authenticity of documents. Information resources can only be managed in an integrated manner without any media discontinuities by providing services for the entire information resource life-cycle. DMS include electronic as well as paperbased information resources that can be treated in the same way as electronic documents once they are captured. Annotation with meta-data is an important means in order to store the document’s creation and application context, track changes and administer different versions. Traditionally, the advantage of DMS was noted that handling of paper-based documents is costly and less flexible. This argument is now complemented by the fact that now, an increasing amount of information resources needs to be maintained electronically over their whole life-cycle. The term content management system sometimes is used synonymously to DMS but in the narrow sense denotes Web content management systems used to publish information on Intranets or on the Web (section 4.4.4). Potential advantages of DMS for an organization are decentralized and simultaneous access by multiple users, flexible categorization and filing of documents, higher consistency through centralized data management and easy backup. Quantifiable advantages in terms of costs and time are faster access, archiving, exchange, search and retrieval of documents, improved productivity of document-related tasks and cost reductions in terms of personnel costs as well as material and archiving expenses. From the user’s perspective, DMS provide dedicated interfaces and clients for specialized tasks, e.g., for capturing documents or administering the DMS. Furthermore, they can be integrated into the user’s desktop with the help of the WebDAV protocol (section 5.2.4), with TIFF or PDF printers that allow to generate application independent formats over the printer interface as well as by integration into end-user applications by means of standardized interfaces. A typical example with a focus on integration into office suites is the ODMA standard (open document management API). It was defined by the ODMA group, an international association of about 50 DMS vendors, and establishes a set of software interfaces that a DMS needs to provide in

Analysis of definition

Advantages of DMS

Interfaces and integration

ODMA

312

Product examples

4. Knowledge Services

order to allow direct access from within end-user applications, e.g., opening and saving a document from within the text processing software. The main advantage of ODMA compared to proprietary solutions is its independency from concrete DMS products. Examples for DMS are IBM Filenet, EMC documentum and Open Text Livelink. During the last years, the DMS and WCMS market was characterized by intensive changes. Many product vendors vanished from the market or were acquired by other organizations. Today, it becomes increasingly important not only to provide a specialized solution, e.g., a product that is able to efficiently and flexibly publish contents on the Web, but rather to offer an enterprise-wide platform that is able to integrate contents stored in multiple different repositories, e.g., file servers, SAP R/3 modules, Lotus Notes databases, Microsoft Sharepoint sites, and to offer means that enable to manage them in compliance to increasingly relevant legal regulations.

4.4.2 Capturing

Possibilities for capturing information resources

Capturing WCMS content

The capturing of information resources comprises their digitalization and storage within the repository of a DMS. Principally, this can be accomplished by entering information resources manually, e.g., in free-text or with the help of forms, scanning paper-based information resources or microfilms with scanners, if necessary coupled with automatic transformation on the basis of optical character recognition, photographing or filming information resources with cameras, digitizing audio with soundcards, capturing and transforming video sequences with video boards, drawing maps, technical plans or engineering drawings with specialized digitizers and migrating existing electronic information resources from legacy systems. Capturing contents within a WCMS (section 4.4.4) can be done by means of Web-based forms where authors can enter text, upload images and add meta-data in a structured way. They can be structured according to pre-defined content types to be captured, e.g., news, articles and organizational unit description, which allows a clear separation between content, structure and layout. Some systems present the resulting Web pages in a pre-view mode with edit-buttons where changes are possible so that authors can type directly into a Web site. Alternatively, WCMS content can be created with other applications such as text processors based on templates.

4.4. In-form-ation Services

313

Paper-based documents are digitized with the help of scanners. In this relation, a centralized scanning approach can be distinguished from a decentralized approach. Centralized scanning denotes a setting where up to several thousands of documents per day are scanned in a centralized location with specialized high-volume scanners. For example, the internal post department scans all incoming mail and then distributes it electronically to other organizational units. In the context of a decentralized scanning approach typically a significantly lower number of paper-based documents is scanned at decentralized workplaces that are responsible for capturing only certain document types, e.g., job applications or specific types of invoices. The relevance for determining whether specific types of paper-based documents should or should not be included in the scanning process can be determined with the help of the following criteria: 1. The general importance of certain document types, particularly the degree of their relevance for organizational business processes, 2. the access rate, i.e. how often a document needs to be accessed by several users, 3. and last but not least whether legal regulations demand long-term storage of specific types of documents. Table 4-6 lists a number of classes that can be defined based on these three criteria. This classification scheme is especially useful during the introduction of a DMS. At this stage, usually large amounts of paper-based documents and micro-films need to be captured and archived. It might be useful to consign companies specialized on high-volume scanning, basic categorization and eventually the destruction of paper-based documents.

Centralized vs. decentralized scanning

Table 4-6.

Classes of documents

class

characteristics

decision

A

documents with high importance and high access rates

store within DMS

B

high importance, low access rates

C

low importance, high access rates

D

low importance, low access rates, but archiving legally prescribed

balance legal restrictions, necessity of safe-keeping outside DMS (paper-based archive), possibility to destroy original documents, and efforts for capturing

E

low importance, low access rates, no legal restrictions

shelve outside DMS

Criteria for document classification

Document classes

314

Early vs. late capturing

Coded information (CI) and non-coded information (NCI)

4. Knowledge Services

Documents belonging to class A definitely need to be archived as they are either very important or have high access rates, or both. For the documents that fall into the classes B, C and D, the efforts for capturing documents need to be balanced with the benefits that can be achieved by storing them within a DMS. The actual inclusion of a document type within the DMS depends on the individual case, i.e. how high access rates actually are, how important a document really is and what the exact requirements of regulations are. Class E are not relevant for the scanning process and thus should be shelved outside the DMS - if archiving those documents is required at all. It is possible that some documents need to be archived in a paper-based format in addition to their scanned versions, e.g., because the authenticity of signed contracts or certificates cannot be ensured otherwise. Nevertheless, DMS are able to provide an easy and fast access to such documents. Documents can either be captured early when they are created or arrive in the organization, e.g., at the time when they arrive by mail. In this case they are scanned, indexed and stored in the DMS directly after arrival, which typically requires a centralized scanning scenario. Relevant criterion for early capturing however is not the location where documents are scanned but rather that all document-related tasks such as reading, making annotations and also revisions are accomplished on the electronic document by means of DMS functions. Late capturing in contrast moves documents into the DMS when all document-related tasks have been done on the paper-based document, typically annotation by means of various stamps as well as placing signatures or comments on the document. This is a typical archiving scenario. Scanning of documents can be done decentrally or centrally. The conformity of electronic documents with paperbased documents can be certified by means of electronic signatures, i.e. algorithms that ensure that no changes have been done on a document as long as the associated secure key is valid. Within the DMS context, it is traditionally distinguished between coded information (CI) and non-coded information (NCI), based on the principal approach of technical representation of documents. CI is structured in a way that enables the direct interpretation of, e.g., characters, words and paragraphs in the case of text documents. Common examples are plain text, rtf, HTML and XML files as well as the file formats of office suites. The term NCI is typically used for image file formats that in contrast to CI cannot be interpreted straightforwardly by a computer. Typical examples are formats such as Bitmaps, TIFF, GIF or PNG. Though a scanned document can directly be read by humans, its single characters cannot be interpreted by a computer as it still is an image of a document. Nevertheless, it is possible to convert some NCI documents into CI, e.g., by applying OCR. A special case are PDF and postscript documents that can be CI, e.g., when created from a text document with Adobe Acrobat, but can also

4.4. In-form-ation Services

be NCI, e.g., when created with a scan software by just packaging the image files coming from the scanner into a PDF container. CI has the advantage that it requires less space than NCI. A typical page requires between 2.5 and 30 Kilobytes whereas an uncompressed NCI requires between 0.5 and 8.3 Megabytes. Furthermore, CI can be processed further, e.g., by the indexing mechanism of a search engine, whereas NCI is typically not accessible by full-text search. The main advantage of NCI is its simplicity. Standardized image formats are easy to visualize without any specialized application. Layout is preserved, which usually cannot be guaranteed by CI. Proprietary file formats might be converted to simple NCI formats because it cannot be ensured that these formats are accessible over the whole prescribed retention period. For instance, if a document was written and stored with Lotus AmiPro in the year 1997, over ten years later it is unlikely that it can straightforwardly opened because no computer has this application installed within the organization or due to the lack of an appropriate document viewer that can represent the document in its original form. Hence, the need to prove that a document is represented in an exact and unchanged way compared to the version originally created can be another reason for the choice of NCI as internal representation. However, the relevance of NCI decreased significantly since PDF/A has been introduced as a standard (section 4.4.5). PDF/A is a recent standard (ISO 19005-1) of a file format targeted at long term readability of documents. It has been declared valid in fall 2005. As a CI standard, it offers richer means to store and describe machinereadable text-documents and thus represents an alternative to the NCI standards TIFF and JPEG. In contrast to other CI standards, it is universally applicable for storing different formats, e.g., CAD drawings as well as scanned images. It is based on version 1.4 of the PDF file format. A PDF/A file is independent from hardware devices, software platforms and different software versions, self-contained, i.e. it contains all data required for presentation such as fonts used in the documents, self-documented, it includes a set of standardized information that describes the document, transparent, it can be analyzed by means of simple technical tools. The standard is subdivided into two levels: PDF/A-1b describes the minimum requirements for documents that conform to the PDF/A standard. It ensures that document can be displayed correctly but does not secure that the text contained is readable and understandable. PDF/A-1a therefore adds Unicode support (section 3.1.2) and “tagged PDF”, which particularly supports accessibility by including semantic descriptions of the text contained and its logical structure. These are also required if the text has to be re-formatted, e.g., in order to be displayed on a mobile

315

CI and NCI compared

PDF/A

316

Optical character recognition (OCR)

Intelligent character recognition (ICR)

Visual processing

4. Knowledge Services

device with a smaller screen. For the near future, it is planned to release PDF/A-2 as ISO 32000, which includes evolutionary enhancements such as support of layers, JPEG2000 support and visual transparency. Optical character recognition (OCR) denotes a set of methods that deal with the transformation of scanned document images into machine-editable text, in other words with the conversion of NCI to CI. This is achieved by structuring an image into many small parts that represent single alpha-numeric characters. These parts then are analyzed with the goal to identify the character that it represents. Since format and quality of the scanned image may vary, e.g., due to different colors, scanning resolutions, font types and layout elements on the page, the analysis needs to be based on a set of “ideal” patterns that allow to estimate probabilities to which a defined part of the image represents a specific character. Early OCR systems required the use of specifically designed fonts, e.g., OCR-A and OCR-H based on DIN 660088 and 66225 respectively, that allowed distinguishing single characters clearly from each other or they required a learning phase with a defined set of training pages. Today, OCR of typewritten text can achieve very good recognition rates that exceed 99% for many different fonts and without special training. More challenging is the recognition of handwritten text where single characters cannot be clearly separated from each other. Recent research concentrates on this topic as handwritten text can be a good substitute for keyboards, especially for mobile devices such as PDAs and laptops. Even a very low number of errors on a page should be avoided as this may require time-consuming manual proof-reading and error-corrections. Therefore, another set of methods is designed to correct OCR results after the actual pattern-matching process which is often subsumed under intelligent character recognition (ICR). For example, OCR may recognize the name “Bill” as “8ill” since “B” and “8” are similar with respect to shape. This can be corrected by comparing the recognized words with those in dictionaries or by applying statistical methods that estimate the probability of correctness based on linguistic approaches or experience values. Visual processing tries enhancing the quality of the scanned images before the OCR methods are applied or the images are automatically processed otherwise. Typical challenges that might prevent the correct recognition of characters at all are low contrasts, text with many broken characters such as dot matrix text, carbon copy text and handwriting, stains, highlighted or underlined areas as well as shaded backgrounds. Thus, visual post-processing optimizes scanned images by various filters that enhance low contrasts, enhance broken characters and suppress backgrounds. This can dramatically increase OCR quality and the ability to automatically process documents further. The quality of OCR results is determined by colors, contrast, layout and font type of the paper-based document, the resolution and quality of the scanned image,

4.4. In-form-ation Services

the quality of the separation of the document layout from original text, the quality of the OCR approach itself, particularly the comprehensiveness and quality of the patterns used for the pattern matching process, the quality of the ICR process, i.e. the quality of the ICR algorithms, suitability of the dictionaries applied and the appropriateness of statistical methods. Document classification deals with the categorization of documents in order to assign them to different roles or business processes. It is a typical task that has to be accomplished by a central internal post office where incoming documents need to be routed to the correct departments. The automatization of this task in relation to a centralized scanning approach may enhance the efficiency of this process substantially, especially when a large amount of well-known documents arrives in the organization, i.e. documents where layout and type of information contained is known. Automatic classification can be based on a training stage where the system generalizes from typical examples of documents that are assigned to a class. When documents of this type are scanned, they then can be automatically classified and by this, e.g., put into selected folders, distributed by workflow or email to responsible roles or be automatically processed further. Ideally, the exact position of specific information or values is known so that they can be made machine-readable by OCR and assigned to predefined data fields. This reduces manual work and enables largely automated processes, e.g., processing of incoming invoices that are scanned, stored within the organizational Enterprise Report Management (section 4.4.5) software (and treated by workflows that present all required information to responsible roles, e.g., in the finance department. However, this requires a careful identification of the relevant documents as well as the continuous adoption of the system to changes as otherwise, the need for manual corrections may outweigh the advantages gained by automation. Barcodes are a widely-used solution that reduces the effort of capturing meta-data about documents. They mainly represent an identification number that may be associated with various different semantics such as document type, customer ID, workflow step or previous versions of the document stored within organizational systems. Principally, they are structured in the same way as the barcodes on common consumer products used, e.g., in order to identify product group and price at the supermarket cash desk. Barcodes are suited particularly for return documents formerly distributed by the organization, e.g., contracts to be signed and service request forms to be filled out. They can be recognized before scanning by means of barcode readers or after scanning based on software algorithms. During a DMS introduction project, it is often required to transfer potentially large amounts of legacy data stored in other systems, e.g., file servers, custom built systems or old DMS products that should be dis-

317

Document classification

Extraction of specific fields

Barcodes

Bulk loading

318

4. Knowledge Services

placed by a new solution. The complexity of this migration process depends on the structure and amount of documents and meta-data. A typical migration project consists of the following stages: Firstly, the data to be imported needs to be analyzed in terms of relevant meta-data, storage structure, types of files and access privileges. Afterwards, a way and structure to export the data from the old system needs to be found. This is often done via an export to the file systems in a target format that can be read by the new DMS system. This information then is loaded into the new system by means of specialized tools or software components that are able to interpret the meta-data, create a new object within the DMS and associate it with the meta-data as defined within the file. Depending on the amount of data to be imported and the available infrastructure, this process can take up to several days.

4.4.3 Manipulation

Access privileges Table 4-7.

Information resources can be of various types, e.g., contracts, presentations, meeting minutes and articles, and are stored in different formats, e.g., PDF, TIFF, OpenDocument and proprietary formats such as those of Microsoft Office, as already noted in relation to the distinction between NCI and CI (section 4.4.2). Hence, various specialized editors and applications are required in order to edit them. EKI manipulation services represent a generalized group of services that can be applied for all those different types and formats. Their focus is not on editing actual information resources themselves but rather on controlling and easing manipulation processes. This is accomplished by providing means to control access privileges, versioning, audit and check-out services. These services are described in more detail in the following. Access privileges control actions that can be performed on information resources. Table 4-7 lists and describes a set of general access privileges. Types of access privileges

Type

Description: User is allowed to ...

read

see meta-data of item and its contents which also may relate to items in a folder.

modify

change associated meta-data such as name and description of an item.

add

contribute new items, e.g., uploading files or adding new versions to existing ones.

delete

remove an item or single versions of it.

edit privileges

define permissions for users and groups which provides full access to the item.

Their implementation, e.g., in DMS products, may vary depending on the actual requirements. For example it is possible to subsume the privileges listed under the groups read, i.e. only read privileges, write, i.e. read,

4.4. In-form-ation Services

modify and add privileges, and administer, which includes all privileges listed in the table. Assignment of permissions requires careful planning as otherwise, the permission structure may evolve very complex and users loose oversight who actually has access to a specific item. Therefore, this is a typical task as part of a DMS introduction project. Figure 4-20 gives an example of how privileges are assigned in the DMS Open Text Livelink.

Figure 4-20. Assignment of privileges to users and groups in Livelink

Though they principally can be changed for single information resources, they are typically assigned to items that contain other items such as folders, in this case the folder “meeting minutes”. The reason is that all items placed within a container inherit the respective privileges, which substantially eases administration. On the left-hand side of the screenshot, the users and groups that have access are listed, separated into a group “default access”, i.e. owner, owner group and public access, and the group “assigned access” for all further groups. Default access comprises standard users and groups that always exist for every item and are changed based on the contributor, i.e. if another user than “Hädrich, Thomas” places an item into the folder meeting minutes then the owner and the owner group are changed. The idea behind this is that access privileges are treated differently for the creators of information resources that commonly have full access and other users that normally have more restricted

319

320

Versioning

4. Knowledge Services

permissions. On the right-hand side of the screen, privileges can be defined for the respective group selected. In this case, the user group “Consulting NORDIC” is assigned “see” and “see contents” privileges which corresponds to the “read” privilege as described in Table 4-7. A version is a particular form or variant of contents or information resources. In the context of in-form-ation services, versions are created for structuring the creation process of information resources, e.g., a meeting protocol where the authors step-by-step add their contributions, maintaining the same contents in different languages, e.g., contents to be displayed on a Web page that can be displayed in English and German, offering the same content in different variants, e.g., a slide set for internal use that contains technical details and a shortened version for external use, providing the same content by means of different media, e.g., a news article can be displayed in a Web browser and stored locally as a PDF file. Versioning mechanisms support the administration of the different associated versions, which includes consistent management of several versions belonging to one document and automatic numbering of versions. Figure 4-21 shows the versioning mechanism of Open Text Livelink as an example. The item “2008-08-01 Meeting protocol” is represented by four versions. By default, different versions either are numbered sequentially or based on a minor/major schema. In case of the example, three authors contributed the minor versions 0.1 to 0.3. Later on, one author promoted the last version to the major version 1.0 which represents the commonly accepted and final version of the meeting protocol. Modern DMS furthermore offer customizable numbering schemes if this is required.

Figure 4-21. Versioning mechanism of Livelink

4.4. In-form-ation Services

Manipulation services also allow to keep track of the actions performed such as creation of items, addition of new versions, changes on meta-data or just read access to specific items. This so-called audit information can be helpful for various purposes, e.g., coordination between co-authors that are able to see who currently has reserved an item, monitoring overall systems usage or the use of specific information types as well as evaluating who has performed specific actions such as the deletion of items or changes on its meta-data. Nevertheless, gathering of information about user actions is a sensitive area. A compromise needs to be found between collecting potentially valuable information and protecting users from illegitimate observation. In practice, a good compromise is to exclude all read access actions from the audit and to only generate reports based on audit information that do not reveal the identity of single users. Check-in / check-out services help to coordinate write access by different users. Check-out means that a person restricts write access, e.g., the privilege to add a new version of the information resource, to his user or group of users defined during the check-out process. From this point, other users can only retrieve the information resource but are not able to do any changes. This prevents simultaneous and thus potentially conflicting changes on information resources by multiple users. The check-in process typically consists of two steps: the addition of a new version as well as the release of the item.

321 Audit

Check-in / check-out

4.4.4 Translation Translation services deal with the transformation of content into different structures and formats. Most important application area for translation services is the publishing of information on the Intranet, Extranet or the Internet. This is supported by the specialized system class of Web content management systems (WCMS) which will be defined and discussed in detail. Web content management systems provide services to efficiently handle all electronic information resources required to design, run and maintain a Web site and to support all tasks related to authoring, acquisition, review, transformation, storage, publishing and delivering contents in Web-based formats. WCMS offer a repository that stores all semi-structured information resources to be transformed and published as well as services to manage these resources. Therefore, it can be denoted as a specialized class of DMS. In contrast to the holistic orientation of DMS towards managing information resources over their entire life-cycle and supporting all related processes, a WCMS concentrates on acquiring, storing and publishing information in Web-based formats.

Definition of WCMS

322

4. Knowledge Services

Representation of content

Please note the use of the term content. Since content is an abstract concept that cannot be stored itself, it needs to be represented in the form of information resource that has a structure which supports its transformation into other structures and its rendering with alternative layouts. A good example is a XML document with a well-defined schema that is transformed by means of an XSL stylesheet and formatted with the help of CSS (section 3.1.5) as this reflects a clear separation of content, structure and layout within the DMS. However, many current WCMS still do not clearly support a separation of structure and layout and therefore have their limits with regards to a fully flexible translation into different formats. Meta-data is equally important as compared to DMS since it governs the retrieval of content and its automatic transformation and distribution into the right formats and channels, e.g., content might be presented differently and with varying levels of detail depending on the target group or roles as is indicated within the meta-data. Templates control the process of transforming content into the desired layout or format of publication. Two types of templates can be distinguished. A template consists of layout elements, e.g, HTML tags, and program instructions that control where and how parts of the content should be placed on the layouted page. The advantage is that only the templates have to be changed in order to display one and the same content in a different way instead of making changes on potentially large amounts of information resources directly presented to the users. A single Web page is usually rendered with the help of multiple templates in a cascading process top down from a main template that defines the overall structure of the page to atomic templates that render single parts of the page. This transformation process is not limited to creating Web contents but may also produce other formats such as plain text, OpenDocument files or PDF files. The latter can be achieved in a standard way by using XSL-FO (see 3.1.5). Two principal server architectures can be distinguished with regard to how changes on contents to be published are propagated. The first alternative is that all changes are published directly. Therefore, content needs to be rendered dynamically at the time it is accessed. This requires high-performance servers as content needs to be retrieved from the database and rendering templates need to be applied in order to produce the intended structure and layout. This alternative is particularly suited for fast changing content that is accessed by a relatively low number of users. The other alternative is to periodically prepare and export appropriately rendered content to a staging area or server where users can access the content. This allows to separate the actual WCMS from the publishing area so that it can be operated, managed and accessed in a secure environment while access of external users can be restricted to a separate infrastructure. This alternative is suited for relatively stable content that is accessed by a large number of users because no further processing is required at the time of access.

Templates

Staging

4.4. In-form-ation Services

Workflows (section 3.4.2, 225ff) can be used in order to coordinate tasks during authoring and publishing contents. The publishing process is a typical example for a workflow. The author initiates the workflow when he has created new content. It is reviewed by a role responsible for quality assurance and potentially routed back to the author with requests for changes. After authoring and review, the contents are released and published by means of the WCMS.

323 Workflows

Publishing with and without WCMS. The value and advantages of WCMS become very evident when comparing traditional Web publishing without WCMS and Web publishing with WCMS. Table 4-8 contrasts both approaches which are outlined in the following. Table 4-8.

Web publishing with and without WCMS compared (Zschau et al. 2002) traditional Web publishing

Web publishing with WCMS

format

content and layout mixed

content and layout separated

publishers

only Web team or Web master publish content, danger of bottleneck

all employees are able to publish content

effort

publishing effort is high and quickly grows with size of Web site

initial effort for introduction of WCMS is high, publishing effort is low and slowly grows with size of Web site

publishing workflow

manual publishing workflow, e.g., based on emails

automated publishing workflow supported by WCMS

tools

multiple different tools required in order to manage Web site

WCMS provides integrated set of functions, no additional tools necessary

redesign

redesign of Web site is a costly reprogramming process

easy redesign by changing templates

Contents to be published on an organizational Web site are usually authored and illustrated by many different people. Most of them do not possess skills to create complex Web pages formatted according to the organization’s corporate Web design, embed resources, place them on a Web server and link them to other pages of the Web presence. In the case of traditional Web publishing, Web masters are responsible for technical management of the Web site which often includes administration of software as well as hardware. They or specialized Web publishers take over the task of transforming contents to be published into the right format and a consistent layout. The danger is that these employees become a bottleneck if many changes have to be made or a large amount of new content should be published. In order to cope with this situation, authors wanting their contents to appear as soon as possible thus are inclined to acquire technical skills to create content on their own. This in turn has the danger that the consistent appearance of the Web site is broken because they will

Traditional Web publishing

324

Web publishing with WCMS

First generation

Second generation

4. Knowledge Services

explicitly or implicitly develop their own ideas about layout and formats. The management of the Web site involves many different tools, e.g., converters for image formats, link-checkers and content validators. Publishing Web content is an error-prone process, especially if the Web presence grows up to many hundred pages. Major changes, especially applying a new layout to refresh the appearance of the Web site are costly and timeconsuming and regularly have to be accomplished by specialized companies. In the case of Web publishing with WCMS, a centralized system enables every author to publish content directly without the help of Web masters or Web designers. Thus, employees with skills to transform and publish the content are not a bottleneck anymore. Layout information is separated from content and shifted to templates that define how contents should be transformed. Changing the site’s appearance is merely a matter of altering layout templates. Consistency of the Web site’s layout is higher and publishing is a less error-prone process because the system has built-in functions to validate results, e.g., broken links are avoided due to evaluation of links by the WCMS. Therefore, some of the tasks formerly done manually are automated and the effort to maintain even large Web sites remains low, growing only slowly with increasing numbers of pages. Publishing and access to contents is coordinated by workflows and user privileges. Overall, time to publish contents on a Web site with a better quality is substantially shortened. Generations of WCMS. The wide-spread use of the WWW as information platform can be dated back to the early 1990s. Soon, it became clear that creation of Web content needs to be supported by appropriate tools. Table 4-9 gives an overview over the four generations of WCMS that can be identified. The first generation includes the precursors of WCMS which are tools targeted at making Web site administration and structuring more efficient. The focus is on easing publication of larger amounts of contents and on introducing regulations and functions for structuring Web sites. Relational databases are filled with content and are used to automatically generate HTML pages. Authors needed to write pages in HTML, hence layout is not standardized neither separated from content. Systems are usually implemented in server-side script languages like Perl. Access to Web contents is administered with the help of Web server functions that usually do not offer functionality for controlling privileges of authors or for managing additional meta-data. The second generation of WCMS introduced many functions that aim at making the publishing and transformation processes of Web contents more efficient, e.g., functions to automatically transform content to Web formats, generate elements for navigation through a Web site, e.g., tables of content and site maps, and to guarantee consistency of the Web site, e.g.,

4.4. In-form-ation Services

325

by automatic link checking. Through provision of editors that are able to generate Web pages, technical Web publishing skills were not necessarily needed anymore and more authors are able to create and publish information. Functions to administer user privileges and built-in publishing workflows to make publishing more effective are now supported by WCMS. Pre-defined meta-data can be stored in order to structure Web site content and to automate content handling, e.g., by storing information about expiry dates removing outdated contents automatically. Table 4-9.

Generations of WCMS (based on Zschau et al. 2002) first generation (mid 90s)

second generation (early ’00s)

third generation (mid ’00s)

fourth generation (outlook)

Web site administration

Web publishing

dynamic Web publishing

enterprise content management

target group

technical experts manage Web site

non-technical authors are able to publish Web contents

WCMS open to all employees

WCMS open for internal and external information providers and sources

meta-data

no meta-data

pre-defined metadata

customizable meta- semantic meta-data data

navigation

manual structuring of contents

automatic generation of navigation paths

basic personalization and profiling

context-dependent, dynamic navigation

rights man- no user privileges agement

basic user privileges

standardized user management

integration with organization-wide identity management

workflow

manual workflow

pre-defined workflows

customizable workflows

ad-hoc workflows and collaboration functions

flexibility

static contents

basic functions to manage flexible contents

advanced functions to support dynamic content

integration of multiple internal and external information sources

integration

no integration with existing systems

application integra- connectors to most tion with APIs important Web applications

WCMS part of an enterprise knowledge infrastructure

The third generation represents the current technical state of advanced WCMS. Customizable meta-data and workflows support publication processes more flexibly as well as dynamic handling contents. WCMS provide interfaces for integration with other systems and with user desktops. Personalization and user profiles are supported (section 5.1.2, 370ff) and

Third generation

326

Fourth generation

4. Knowledge Services

allow Web site visitors to individualize the appearance of specific page, e.g., the information and functions displayed within an Internet portal. Web publishing is more and more open for all employees of the organization due to enhanced usability, not only for designated authors. The fourth generation gives an impression of how WCMS might develop in the near future. A major trend is integration with other systems that support publishing and collaboration around contents. In particular, WCMS are more and more integrated with DMS to facilitate handling of other formats, content with varying life times and archiving. Standardized interfaces for integration with other systems as well as different information sources from inside and outside the organization are provided. Support of Semantic Web standards (e.g., RDF(S), OWL, section 3.3, 205ff) and enriched meta-data is a foundation for semantic links to other Web contents and for the flexible handling of large amounts of contents. Advanced functions for user profiling as well as enhanced categorization and personalization enable context-dependent and dynamic navigation. WCMS are also able to act as a communication platform by provision of modern communication services such as wiki books and Blogs and publishing multimedia contents based on Flash player or similar technologies.

4.4.5 Archiving Archiving concerns efficient long-term storage of information resources on appropriate storage media and management of retention periods. This section firstly discusses the services relevant, turns to standards that have been formed in this area and then summarizes respective legal aspects.

Virtual storage

Services. Important archiving services are virtual storage, single instance storage, controlled deletion, tamper-proof storage and COLD / ERM, which will be outlined in the following. Furthermore, this section described email archiving as one of the recent challenges in this area. Archiving systems need to offer a virtualization layer that abstracts from actual storage media and systems. On a file system level, the files archived can be either stored separately or are aggregated within an archive, e.g., a so-called ISO file that represents a DVD image which may include compression of files in order to reduce the storage space required. The choice of the best alternative depends on the file size (smaller file size argues for aggregation due to smaller storage overhead) and on the question whether single files need to be changed or deleted, which argues for individual storage as otherwise the whole archive file needs to be reorganized. However, the actual access by system users should not be affected as the virtualization layer abstracts from the concrete way of storage including the storage media actually applied. Virtualization makes it also transparent whether the files are stored on faster and more expensive, or slower and cheaper storage media (see HSM in section 2.3.1).

4.4. In-form-ation Services

327

Single instance storage refers to the ability of the system to keep only one copy of information resources that multiple users or computers share. It is typically implemented on email servers and archiving systems. On email servers for example, one and the same email is stored only once while it can be accessed through several inboxes of the receivers by means of a reference. In relation to archiving it is used to reduce the amount of required storage space as it avoids the accumulation of redundant copies of the same information resource. The process of crawling archives in order to identify and remove redundant copies is also referred to as deduplication. Though single instance storage is the result of this process, both terms are sometimes used synonymously. Advanced archiving systems also support single instance storage across source systems, so that a file is identified as identical and stored only once no matter whether it is an attachment of an email, a file on a file server or a document in an DMS. Organizations are required to archive information resources only for a limited amount of time. Therefore, an archiving system needs to offer services for the identification and deletion of items due for deletion. Though freeing storage space also is a reason for this, most important is that every document represents a potential piece of evidence that can be used against the organization, e.g., in the context of a legal case. Hence, destroying information in time also reduces legal and thus financial risks. One way to implement this are retention searches. These are based on meta-data assigned to information resources at the time of their storage, e.g., type and date of last modification as well as on retention schedules that control the amount of time each type of resource has to be stored. An important requirement that archiving services face is the offering of a secure storage space that ensures that none of the files stored can be changed. One way to implement this is the application of storage media that can only be written once such as WORM. Another possibility is the use of digital signatures that indicate whether a resource has been changed and thus can be used to prove the originality of an electronic item (see below). Today, companies such as NetApp, EMC or Hitachi Data Systems offer hard drive-based products that simulate a write-once behavior for defined retention periods. The data stored on these cannot be deleted or changed as it is secured by an internal operating system. COLD (computer output to laser disc) deals with capturing and archiving of reports such as loan records, customer bills and accounting reports that should be archived in parallel to print-out on storage media such as hard drives, storage area networks or optical media. It usually concerns large reports or print-jobs and was typically introduced to replace paper-based or microfiche solutions. The term COLD more recently has been superseded by Enterprise Report Management (ERM). Typically, the data is generated as print or report job and transferred to the COLD software component, which decomposes the data stream into logical units

Single instance storage

Controlled deletion

Tamper-proof storage

COLD, ERM

328

Email archiving

Compliance

Trade and tax law

4. Knowledge Services

such as single documents or pages, extracts indexing information and stores the data within the archive. More and more organizations particularly in the pharmaceutical and banking industries face the obligation to archive all incoming and outgoing emails as they may contain legally relevant information. Challenges stem from the huge amount of data that archiving systems have to cope with. It is not unusual that several thousand email boxes need to be archived which may generate several GByte of data per day (as an example, a company with 8,000 email boxes uses 38 TB for archiving emails per year). The problem is not only that this data needs to be compressed and written to secure storage media but also that every email including attachments needs to be indexed in order to be retrievable by means of structured search queries. Usually, relevant emails are identified by keyword-based searches and the corresponding results are transferred to lawyers that analyze the results step-by-step. Therefore, the quality of the index is crucial, particularly the inclusion of all relevant emails. Legal regulations. Handling of documents is influenced by a large number of legal regulations. Their relevance for organizations is determined by country, industry, activities and legal form of the company. The term compliance has been coined to denote accordance and fulfillment of legal and regulatory specifications. Accordance refers to clearly defined, official specifications that concretely state what has to be done. This can be evaluated directly by comparing rules and practice. Fulfillment in contrast refers to requirements that some result has to fulfill. Consequently, how to achieve accordance is relatively clear as rules just have to be fulfilled that statically prescribe behavior whereas fulfillment requires a more comprehensive approach that dynamically monitors and controls the processes that create a specific result. The behavior of car drivers for example accords to the rule “the maximum speed is 50 km/h” if they drive not faster than 50 km/h whereas “the roads in the city must be safe for every user” is a requirement that can only be fulfilled by continuous control of and intervention into multiple processes, e.g., the actual traffic, traffic planning and road construction. Consequently, compliance is a part of corporate governance as it requires an overview and control of organization-wide processes. Relevant laws aim at orderly, secure and traceable storing of documents that contain relevant data about business transactions. Most regulations are nation-specific, though the European Union strives for harmonization of national laws in Europe. A comprehensive event such as the SarbanesOxley Act is missing in Europe. As an example, the laws that are significant in Germany are outlined here (Glubins et al. 2002). Trade and tax law, i.e. Handelsgesetzbuch (HGB), Grundsätze ordnungsgemäßer Buchführung (GoB) and Abgabenordnung (AO), specify general standards and procedures of book-keeping. They regulate the types

4.4. In-form-ation Services

of documents to be archived and the respective time period. Storage on electronic data carriers is explicitly permitted, though the term “data carrier” is not clearly defined. Additionally, general requirements for accounting need to be considered in DM such as accurate and complete recording of all accounting transactions. For audits, an electronic archiving system must be able to present all relevant records with adequate effort, promptly, in the required sequence and for any point of time. Civil and criminal law, i.e. Zivilprozessordnung (ZPO), Bürgerliches Gesetzbuch (BGB), Strafgesetzbuch (StGB) and Strafprozessordnung (StPO), affect DM with respect to the code of civil procedure and in relation to validity of electronic documents for evidence. Law explicitly demands original documents, e.g., financial statements, annual accounts and invoices since only these documents are suited as valid evidence. Thus, DM has to ensure and make verifiable that the electronic document matches the original paper-based document and that changes on the electronic document are always detectable. Electronic signatures are one way to fulfill these requirements. According to the Uniform Electronic Transactions Act (UETA) an electronic signature means “an electronic sound, symbol, or process attached to or logically associated with a record and executed or adopted by a person with the intent to sign the record.“36 It is a concept used in law and needs to be distinguished from digital signatures which are the output of a cryptographic algorithm that has been applied to some data, e.g., in order to generate a checksum or digital fingerprint (see section 2.3.2). In electronic environments, digital signatures are often used to incorporate electronic signatures in order to assure message integrity and authenticity. Signature law (in German: Signaturgesetz) aims at creating the basic conditions for electronic signatures and defines organizational and technical requirements for certification authorities that issue qualified electronic signatures and provide the constituting public key infrastructure (PKI, see section 2.3.2). Other important regulations are data protection laws (in German: Bundesdatenschutzgesetz, BDSG) for data security and protection of personal data, copyright law (in German: Urheberrechtsgesetz, UrhG) for documents secured by copyright, the works constitutions act (in German: Betriebsverfassungsgesetz) particularly during the introduction of a DMS as well as further laws relevant for industries, e.g., emission laws that determine documentation that has to be in place for machinery in plants.

36

The UETA is a Uniform Act drafted by the National Conference of Commissioners on Uniform State Laws (NCCUSL) with the goal to create uniformity between states and draft acts. URL: http://www.law.upenn.edu/bll/archives/ulc/fnact99/1990s/ueta99.htm, last access: 2008-10-30

329

Civil and criminal law, signature law

Electronic signatures

Data protection law, copyright law

330

4. Knowledge Services

Basel II

An international regulation that influences archiving is Basel II. It regulates the conditions of borrowing capital from financial institutions. For that, organizations need to undergo a rating that is generated on the basis of key financial data. This data is stored in accounting systems, management information systems and DMS. The association of organizational information systems (Verband Organisations- und Informationssysteme), an alliance of leading DM vendors, has developed guidelines for archiving conformant to tax and trade laws. Ten maxims guide handling of documents according to the principles of electronic accountancy systems which are called code of practice. The principles clarify the fundamentals that have to be observed in relation to legal laws and regulations when implementing a DMS: 1. Every document that is archived must remain unchanged. 2. Every document must not get lost in the archive or on its way to it. 3. Every document must be locatable with adequate retrieval techniques. 4. Exactly that document must be found that has been searched for. 5. A document must not get destroyed over its entire life time. 6. Every document needs to be displayed and printed in exactly the same format as it has been captured (preserve layout). 7. Every document has to be recovered promptly. 8. All actions that change the organization and structure of the archive have to be logged so that its original state can be reconstructed. 9. Construction of electronic archives has to enable migration to new platforms, media, software versions and components without loss of information. 10.The system has to provide functions that secure compliance to laws as well as organizational regulations regarding data security and life time of the archive.

Code of practice

Standards. The legal regulations that organizational processes need to comply with are not concrete in terms of the features of the technical systems to be used. The concrete requirements need to be derived and broken down based on the relevant rules. Within the area of document and records management, a set of standards has been defined that prescribe how actual systems need to look like. Therefore, several national and international standards have been formed that formulate more or less concrete requirements for record keeping systems. The certification of software based on these standards helps organizations to select software systems that afford their requirements. Most of those standards are relevant for the public sector. Depending on actual industry, other standards can be important such as DIN quality assurance norms for engineering and Basel II in the financial sector. Furthermore, there are many other standards that deal with long term storage,

4.5. Instruction Services

document formats, meta-data, retention periods and retrieval and security mechanisms. Model Requirements for the Management of Electronic Documents and Records 2 (MoReq2) is the most recent specification for software certification. Compared to the other standards, it is the only really international one. It represents a modularized, improved and extended version of MoReq. In fact, MoReq2 is not a standard but a guideline for electronic records management for the public and private sector to support users in system evaluation and implementation and vendors to design such systems. Not included are the actual implementation of the functionality as well as platform-specific and sector-specific requirements. It is structured into a core module that deals with, e.g., classification, retention, administration and several optional modules, e.g., for physical records, workflow, signatures and record keeping processes. The body of the specification is represented by a large number of testable functional requirements to be fulfilled by software used for records management, in terms of this chapter DMS. For example: “The ERMS must be able to support a classification scheme which can represent files and records as being organized in a hierarchy of classes.” These are complemented by short descriptions and explanations. As the fulfillment of the MoReq2 specification will be included in many requests for proposal, the DM industry is working intensively on the certification of their products. Other examples for archiving standards are DOMEA, US DoD 5015.2-STD and VERS.

331

MoReq, MoReq2

4.5 Instruction Services Instruction services represent the final group of knowledge services and target turning material provided in the former steps of the maturity model, particularly formal documents from the in-form-ation step, but also individual contributions from the interaction step, into resources that can be used for self-guided learning or formal instruction. Ultimate aim is to didactically refine material in order to help participants building individual and team knowledge as well as developing skills. Although instruction services generally can be applied for all types of knowledge, they are primarily instruments to facilitate communication and to transfer implicit, personal knowledge. Figure 4-22 shows four main categories used to subdivide instruction services which provide the structure of this section and are briefly explained in the following. Contents produced in the preceding steps of the maturing model are typically not well-suited for supporting self-guided learning or formal training. Enrichment services help knowledge workers to refine material deemed suitable for learning with the help of pedagogical and didactical

From enrichment to assessment

332

4. Knowledge Services

approaches. Composition services use prepared learning material or learning objects as input for creating courses, i.e. arrangements of related learning material or objects. Consumption marks the transition between design time of learning resources and run-time. It primarily consists of using courses delivered to learners by a run-time environment such as a learning management system which needs to be administrated. Finally, assignment services describe a plethora of test and examination instruments that can be used in order to evaluate what has been learned by the participant. enrichment

assessment

composition

consumption

Figure 4-22. Overview of instruction services Role of EKI

Section overview

An EKI integrates instruction services with the other knowledge services on the basis of integration services that allow connecting contents targeted by investigation, individuation, interaction, in-form-ation and instruction services. An EKI also needs to support informal learning, i.e., a process taking place in everyday experience, often on subconscious levels. Research suggests introducing informal learning not as a replacement for formal learning activities, but as a complement to them. Workplace learning is formal and informal learning at users’ work environments. In professional organizations, a mentoring relationship between experienced individuals and novices is often an important source of informal learning for both, the expert and the novice. At the beginning of this section, we will introduce some foundations that are important for understanding different views on learning as well as different types of learners and how this influences decisions about which techniques are best to support and enable learning (section 4.5.1). Then, services that aim at supporting the design of learning modules (section 4.5.2) as well as courses (section 4.5.3) and related issues are discussed. Learning objects are presented as examples of what such learning units, modules or courses could look like in the context of an EKI. Finally, we will deal with services and functions of e-learning suites that support deployment of learning material for use by learners and coaching by trainers (section 4.5.4) and with functions that aim at supporting examinations and tests to give feedback about learning progress (section 4.5.5).

4.5. Instruction Services

333

Learning, like knowledge, is a concept that has a long tradition in academic disciplines such as psychology or pedagogy. The field of knowledge management which provides the basic concepts for the other types of knowledge services from investigation to in-form-ation and the field of elearning which caters for instruction services have been around for more than a decade. Despite numerous attempts to bridge the gap between the two intuitively strongly related fields, they are still quite separated in research and in practice. In large organizations, e-learning and knowledge management are institutionalized in different organizational units, e.g., human resource management versus quality management, different information systems, e.g., learning management systems versus document, content or knowledge management systems and, last but not least, in different attitudes towards the handling of knowledge, e.g., a more central approach targeting formal training and education with pre-defined courses versus a more decentral approach targeting harvesting of knowledge in projects and directly handing it on to knowledge seekers without much effort put into validating and didactically refining the material as well as examining success of learning processes. During the last decade, knowledge management has been extended by concepts, though, in order to cover a broader area that considers the notion of learning and knowledge as being two sides of the same coin. e-learning and knowledge management have in common that they support collaboration or collaborative learning which can provide one starting point for their integration. Another starting point can be found in documented knowledge. The main difference between knowledge conveyed in learning material, e.g., learning objects or courses, versus lessons learned or best practices can be best described as being of varying stages of maturity: from inform-ation to instruction. Chunking courses into learning objects moves them closer to knowledge elements commonly handled in EKIs. In addition to the characteristics of knowledge elements, learning objects need to take into account didactical aspects and make explicit the approach taken towards learning, such as the employed learning paradigm or the targeted types of learners which will be described in more detail in the following.

(E-) Learning and knowledge (management)

4.5.1 Foundation

Learning paradigms. Due to different views and perspectives on what exactly learning is, how it happens and by which means and ways it can be supported, it is not easy to give a commonly accepted definition of learning. A general definition is that learning comprises all changes of behavior caused by experiences. We will shortly outline and compare three different learning paradigms that result in different views on how learning can be supported and how e-learning software should be designed.

Differences ...

... and commonalities

334

4. Knowledge Services

Behaviorism

Behaviorism views learning as reflex under certain conditions. Stimuli evoke corresponding responses of the learner. Ultimately, behaviorism focuses on creation of conditioned reflexes and is often referred to as classic conditioning. It is based on the experiments of Pawlow and was later extended by Skinner’s reinforcement theory which postulates that behavior is more likely to be repeated when it is rewarded. In 1958, Skinner developed the concept of programmed instruction that applies behavioristic assumptions to technical media. It led to the construction of simple mechanic learning machines and later was transferred to learning software. Questions are sequentially presented to learners who have to formulate or choose correct answers. Goal is to generate as many correct answers as possible since only correct behavior reinforces learning. Critiques to behaviorism and learning machines that build on this view state that learners are only motivated to “push the right button” and to reproduce memorized fact knowledge. Thus, behavioristic methods only seem suited to teach basic fact knowledge. Cognitivism focuses on internal thinking processes of learners and therefore differs fundamentally from behaviorism. The approach states that external stimuli are processed autonomously and individuals cannot be directly controlled by them. The brain is seen as an information processing unit that is based on internal dynamic patterns. On an abstract level, brain and computer are compared in this view in that they are both information processors. Cognitivism focuses on problem solving, it is not just about finding the right response to certain stimuli, but to find methods and ways to find good answers to a given problem. Critique states that cognitivism focuses too much on given problems, neglecting the fact that problems often first need to be recognized. Additionally, it is critically pointed out that the human body plays only a minor role in cognitivism so that this view has difficulties with explaining bodily skills. The core assumption of constructivism is that there is not only one objective representation of the world, but that knowledge is constructed autonomously by individuals based on their concepts and ideas. Thus, many different meanings and representations are principally possible for one and the same concept. One “situation” might appear differently for different human actors. Learning is seen as an active process in which knowledge is linked to previous experiences and human actors recognize and solve problems in context-dependent situations. Education thus needs to consider the present knowledge of learners and should involve and activate them. Communication and cooperation play an important role because social interactions are an important means to acquire knowledge. Table 4-10 summarizes the described learning paradigms. The table also characterizes some implications on learning software that considers these paradigms.

Cognitivism

Constructivism

4.5. Instruction Services

Table 4-10.

335

Learning paradigms compared (Baumgartner/Payr 1994 cited from Röder 2003) behaviorism

cognitivism

constructivism

metaphor of brain

passive container

information processing unit

informationally closed system

knowledge

stored

processed

constructed

learning goals

correct answers

good methods to find answers

to master complex situations

learning paradigm

stimulus-response

problem solving

construction

strategy

instruction

observing and helping

cooperation

role of teacher

authority

tutor

coach, trainer

feedback

externally given

externally modelled

internally modelled

interaction

pre-defined and fixed

dynamically dependent on teaching model

self-referring, circular, determined by structure, autonomously

characteristics of software

fixed sequences, quantitative time/answer statistics

dynamically controlled sequences, predefined problems, analysis of answers

dynamic, complex networked systems, no predefined problems

software paradigm

teaching machine

artificial intelligence

social-technical environments

ideal software classes

drill & practice

adaptive systems

simulations, microworlds

Types of learners. Besides different theories and views on learning in general, individuals differ with respect to the way they learn. Each individual prefers a way of learning and for effective knowledge acquisition. This depends on a variety of factors such as motivation to learn, different ways of working with texts, ways of processing information, confidence of the learner in success vs. fear of failure, reactions to difficulties during learning, planning of time and work, attitude towards examinations as well as preferred channels of learning. These factors can be used to define types of learners. Ideally, learning contents are prepared for alternative types of learners so that each learner is able to access learning materials in the format that best fits her needs. However, e-learning must not impose too much additional efforts on designers of contents and thus regularly pragmatic approaches are chosen. Basically, learners can be distinguished to be intrinsically or extrinsically motivated. Intrinsically motivated people learn because they are interested in the learning subject or enjoy dealing with it. Presentation of learning contents that support intrinsically motivated learners should abide

Factors for distinguishing types of learners

Intrinsic vs. extrinsic motivation

336

Preferred media and senses

Learner taxonomy

Divergers

Assimilators

4. Knowledge Services

curiosity and leave control over learning paths to learners. Ideally, learners are be able to dive into a “learning world” with abundant information. Tests and examinations have no primary role and should be presented on demand of the learners. In contrast, extrinsically motivated individuals learn to reach defined goals like in order to earn certificates or diplomas, secure their current workplace, get on their career or raise recognition. Extrinsic motivation leads to different needs with regard to the design of learning materials. They have to be clearly structured into manageable and consistently presented learning units. Motivational measures, tests after presentation of materials, feedback about learning progress and consideration of breaks are more important issues. Active, self-motivated learning is supposed to be more effective than passive consumption of learning contents. Active learning typically is the result of intrinsic motivation which thus should be fostered as much as possible. Another way to distinguish learner types is to categorize learners according to the media and senses they prefer for learning: Auditory user types prefer acoustic presentations, visual user types prefer figurative presentations, abstract user types prefer textual presentations, and audio-visual user types prefer video-based presentations. These groups may directly be regarded when designing learning materials, but do not take into account other important factors like existent knowledge, motivation to learn and learning behavior. Thus, such a categorization is limited with regard to its applicability for individualizing contents. Kolb classifies learners according to two dimensions: (a) according to their preferred learning behavior and (b) according to the way how learners generate new insights (Kolb 1976). He separates two classes of learning behavior. Learners can either prefer to actively experiment or to observe things reflectively. New insights can either be generated by forming abstract concepts or by practical experiences which is close to the differentiation between induction (inference of a generalized conclusion from particular instances) and deduction (conversely to induction the inference in which the conclusion about particulars follows necessarily from general or universal premises). Using these two dimensions, Kolb’s learning style inventory (Table 4-11) distinguishes between four types of learners: Divergers rely on observance and on practical experience, they excel at viewing subjects from different perspectives. They are often very creative and imaginative individuals. Emotions play an important role for them when dealing with ideas. Assimilators prefer to observe reflectively and to form abstract concepts. They like to deal with things and theoretical models rather than with

4.5. Instruction Services

human beings. They are best at constructing coherent theoretical models whose practical applicability is not in the center of interest. Convergers like to experiment actively and to form abstract concepts. A main difference to assimilators is that they strive for practical application of ideas and theories. They focus on manageable situations and clear-cut solutions and are oriented rather technically than emotionally. Accomodators learn through active experiments and practical experiences. They are good at adapting to new situations and inclined to accept risks. They create plans and theories, but fast discard them if they do not fit with new perceptions and insights since forming abstract concepts is not their way of learning. Again, this categorization should be applied with caution. It can only give a rough orientation. Especially in connecting disciplines this model leads to assertions contradicting each other. Table 4-11.

337

Convergers

Accomodators

Kolb’s types of learners (Kolb 1976) preferred learning behavior

way of generating new insights

forming abstract concepts practical experiences

experiment actively

observe reflectively

converger

assimilator

accomodator

diverger

A central factor determining didactic methods and the way to design learning contents is the learner’s expert level. According to Dreyfus/Dreyfus, the level of experience can be classified into novice, advanced beginner, competent, skillful master and expert (see ”Expertise” on page 14). In contrast to novices that rely on descriptive knowledge, expert knowledge and thinking is organized differently and strongly based on experience and intuition. Each level of becoming an expert requires other didactic strategies which are contrasted in Table 4-12. Novices need structured learning material that is presented clearly and manageably. Tests and repetition are best for them to deepen fact knowledge in a certain domain. Advanced beginners can be supported by tutors that answer detailed questions and give feedback about the learning progress. Competents need to discuss and learn about actionable knowledge in groups by jointly working on a task, deepening practical knowledge and opposing different views and perspectives. Complex simulations can be used individually or in moderated group sessions to support learning of skillful masters that already possess knowledge about a variety of typical situations and do not rely on the analysis of single attributes of a situation. Experts only can acquire new knowledge by researching a subject. There are no pre-defined didactic methods as research can take place

Expert level

338

4. Knowledge Services

in a variety of forms. Sharing knowledge and experiences with other experts is equally important. Table 4-12.

Didactic strategies and expert levels (based on Röder 2003)

expert level

Distance learning and telelearning

E-learning

Definition of e-learning

didactic strategy

novice

drill & practice

advanced beginner

support by tutors

competent

work in groups, communication of actionable knowledge

skillful master

explore complex simulations

expert

research, share knowledge with peers

History and trends of computer-supported learning. Distance learning and e-learning are not completely new topics. The roots of distance learning can be traced back to 1865 where first letters with instructions for learning languages were distributed. This was called the “method Toussaint-Langenscheidt” named after it’s inventors. Distance education denotes a process of communicating knowledge between geographically separated teachers and learners. The term distance learning is used synonymously, sometimes to emphasize the interaction between teachers and learners. If communication is supported by electronic media, this activity is characterized as telelearning. The term synchronous telelearning is applied when teachers and learners communicate at the same time over electronic media, e.g., by means of video conferencing. In the other case, when teachers and learners are geographically and temporally separated, one speaks of asynchronous telelearning. In this case, usually material from a storage medium (e.g., a DVD) is accessed. E-learning and technology-enhanced learning are much younger terms that emerged in the late 1990s together with wide-spread use of the Internet and similar terms like e-business or e-government. E-learning is understood as a continuous, life-long process resulting from acting in situations that is supported by ICTs. A more formal training approach with predefined courses contrasts a more informal approach “harvesting” knowledge in projects and directly handing it on to an unspecified target group without much effort for validating, didactically refining or examining success of learning processes. Our definition of e-learning emphasizes that multimedia contents are provided online or offline together with services that enable interaction, though e-learning is often used in a broader sense as comprising other forms of electronically supported learning. E-learning is learning enhanced by ICT with the help of multimedia or hypermedia contents accessible to learners backed by services enabling communication among and between learners, tutors and teachers.

4.5. Instruction Services

A first attempt to support learning with ICT was programmed instruction (See “Behaviorism” on page 334). The goal was to infer the competency level of learners and consequently to individualize learning contents. Learners should be given the possibility to learn autonomously with a system that guides them through the content. Due to restricted access to computer systems in these times, programmed instruction primarily was applied to teach knowledge about programming languages. This approach later was continued under the topic intelligent tutorial systems. This class of systems is characterized by complex models of learners and regularly incorporates techniques of artificial intelligence to intelligently individualize the contents. Programmed instruction was criticized for restricting learners too much to pre-defined learning paths as well as for the underlying behavioristic assumption that learning is solely enforced by positive feedback, which was falsified in scientific experiments. An alternative approach called explorative learning emerged in the 1980s when hypertext and hypermedia systems offered the possibility to freely navigate through contents. Hypertext is any text that contains links to other documents. Words or phrases in the document are linked to other documents that can be easily retrieved and displayed to the reader. Hypermedia extends hypertext with multimedia objects, including sound, video, and interactive 3D animations. When computer systems became technically more advanced, it was possible to create contents and learning systems that needed more storage space and faster PCs. At that time, computer-based training (CBT) emerged which stands for multimedia learning units providing text, sound and video as well as complex interactions between learners and the computer. It was usually distributed on CD-ROM. In the first phases, separation of programmers and authors of learning contents in large projects to create learning modules led to problems. Often, a well-designed technical implementation was emphasized over didactically sophisticated solutions. Later on, authoring tools eased the creation of learning units, though it still remains a time-consuming and expensive process. When Internet and Intranets came into widespread use, Internet technologies were applied to design learning units. Computer-based training became Web-based training (WBT). Optically appealing user interfaces and well-designed contents were emphasized. Unfortunately, the potentials of Web technology to enable communication between learners and teachers as well as among learners was often not fostered. Developers of e-learning systems and its predecessors aim at supporting learning solely through the use of ICT. In contrast, blended learning postulates an integrated, harmonized mix of methods for online, offline and face-to-face learning. For example, participants of university courses prepare at home using online course material and additionally meet in seminar sessions. Blended learning is a more recent term in this area, approximately since 2000. Learning with new media is still in focus, though the

339 Programmed instruction

Hypertext and hypermedia systems

Computerbased training

Web-based training

Blended learning

340

Workplace learning

Professional learning

Advantages

4. Knowledge Services

amount of non-ICT-supported learning sessions and ICT-supported sessions may vary depending on the context. For example, a university could use computer technologies to enhance its pre-existing courses but still rely on seminar sessions. A company that previously offered only virtual courses, could enrich its services by some face-to-face learning sessions. One reason why blended learning is seen as a promising approach is that it combines advantages of traditional face-to-face learning and of new technologies to make learning more effective. Nevertheless, up to now, it does not substantially lower costs of learning, e.g., by systematically reusing courses or bigger target groups, or by establishing bigger audiences. Learning often happens ad-hoc, bound to specific situations and on the job, i.e. during daily tasks. The boundaries between using EKI for searching, retrieving and applying knowledge elements for business tasks and using EKI for learning about new topics are blurry. Thus, workplace learning represents the vision of ICT-supported learning on the job, directly at the desk, in the laboratory or shop floor. Workplace learning aims at flexibly incorporating knowledge elements that are created as by-products of daily operations, especially when exceptions are handled, cooperation with co-workers resulting in learning as well as preparing learning material in a way so that it can be deployed directly at the work place. Backed by EU funded projects and networks of excellence, such as Kaleidoscope, Stellar and especially Prolearn, the concept of professional learning describes a subset of workplace learning targeted at professionals working in industry and services, particularly knowledge workers. In this case, the differences to more traditional learning settings, such as at school or at the university are stressed in order to develop solutions that fulfil the specific requirements of professional knowledge workers. Despite the progress made in the development of e-learning as described above, some challenges remain. Efforts to overcome the product- or push-oriented metaphor inherently part of many applications and embrace the user-centric metaphor are typically part of newer learning approaches. Still, to move towards on-demand presentation of learning material according to the situation in which the user currently is in and to link the resulting learning activities to business processes is achieved seldom. Advantages and disadvantages of e-learning. Many hopes are connected to the application of new media and especially to Internet technologies to support learning. Especially the ability to span geographical and temporal barriers and powerful methods to design contents regularly propel trust in the potenials of e-learning. A major advantage is that learners can freely choose time and place for learning. In countries where it is necessary to travel large distances to the next school or university, the hope is that e-learning could substantially reduce travel costs. Some countries, e.g., Australia, have a long tradition in offering distance educational services. However, companies also could

4.5. Instruction Services

profit from similar cost advantages. E-learning could be a possibility for (corporate) universities to offer their services in a large region and to reduce their costs substantially by reusing contents and improving efficiency of creation and distribution of learning contents as well as of teaching and examinations. Cooperation can be fostered by educational institutions in different places and countries. Digital media offer new possibilities to comprehensibly explain complex issues with the help of animations, simulations, audio and video content as well as virtual (3D) worlds and to personalize learning. Connections to external databases, library catalogues or Web resources enable learners to retrieve additional material and to autonomously deepen their knowledge. Collaboration and cooperation between learners are possible and it is easy to directly contact experts in a special domain which could motivate learners and support their creativity. Altogether, digital media are seen as a way to involve learners more actively. As noted earlier, this is seen as supporting for effective learning. However, e-learning may also have disadvantages. Social interaction between learners over electronic media like chat, email and discussions is not of the same quality as direct face-to-face interaction. Thus, e-learning is criticized of isolating learners rather than fostering cooperation. Many forms of learning, e.g., discussing case studies in group sessions, and some ways of teaching, e.g., demonstrative experiments, cannot be completely substituted by electronic learning contents. Supporting teachers in effectively publishing contents in an e-learning system is a core success factor of e-learning systems.

341

Disadvantages

4.5.2 Enriching Enriching in this case means enhancing media objects, contents or, in the widest sense, digital resources for learning. This includes the definition of learning goals, structuring and arranging contents with learning in mind as well as annotating contents for retrieval, reuse in other domains, automatic detection and selection. A key means for enriching is the translation of documented knowledge into learning objects. Goals are to improve teachability in order to help instructors handing on the material and to improve comprehensibility in order to ease its consumption or re-use. A learning object is an (1) atomic, (2) formalized, (3) self-contained, (4) composable, (5) annotated digital resource (6) that is prepared in a way so that it can be reused to support learning, education or training. (1) Atomic refers to the fact that learning objects are typically small artifacts requiring minutes rather than hours for consumption. However, in order to facilitate learning about a clearly focused topic, a learning object may contain a collection of heterogeneous types of contents, e.g., picture, text and audio. (2) Reusability demands that learning objects are formal-

Definition of learning object

342

Lifetime

Adaptability

Manageable topics

4. Knowledge Services

ized, i.e. represented in a standard format, e.g., XML. (3) Self-contained means that learning objects can be consumed one at a time without the need for further resources to understand the chunk of content. (4) Composable refers to the motivation that learning objects can be composed into larger units such as courses or learning units. (5) Annotated represents the fact that learning objects are described by meta-data typically embedded in the learning objects so that it fulfills the claim of a self-contained digital resource. (6) Due to the aim of reusability, learning objects are prepared so that their characteristics are comprehensively described. In addition to generic meta-data as standardized in the DublinCore (section 3.2.1, 175ff), learning objects are described including specified learning goal(s) that can be reused in potentially predefined contexts and environments. Reusability also might refer to functions of a learning object repository as a container holding learning objects, supporting flexible extension and enhancement. Such a repository ideally is a part of the organizational knowledge base. An important success factor for e-learning in organizations is to provide ways to efficiently design and update learning objects that aim at transferring knowledge relevant for daily operations on the workplace. In contrast to learning objects that aim at transferring knowledge about fundamentals that are stable over a long period of time, this knowledge can be characterized as fast changing. Thus, extensions, updates and enhancements of learning objects need to be brought to learners as quickly and efficiently as possible which are main aims of the concept of learning object. The content of learning objects should be presented in a flexible way to learners according to learner- or learning situation-oriented requirements. The EKI should provide means to regard previous knowledge, skill level and other personal characteristics like interests in specific domains or learning goals, as well as the current work context. Adaptation of contents is distinguished from adaptation of the presentation and navigation of contents. The contents of a learning object thus need to be manually or automatically enriched by meta-data and stored in a repository to flexibly transform, process and thus personalize and publish it for access over different information channels. This is dealt with when composing (section 4.5.3), deploying and assisting with the learning activities (section 4.5.4). Learning objects should be focused on a manageable topic that allows to include meaningful and concrete elements and to make them reusable. If the topic of a learning object is chosen too general, there is the danger of contents remaining too general or the object becoming too big so that learners have no possibility to overview and quickly get a grasp of the topic. The appraisal of what level of granularity is suited depends on the perspective of the actual users. Examples for topics covered by learning objects created by the authors of this book are “success factors and barriers of KM”, “creating skill profiles for KM”, “maintaining electronic community spaces for KM” or “evaluating success of KM”.

4.5. Instruction Services

Learning objects are created, grow, mature, might be frozen or split into more detailed learning objects. This is another example of knowledge maturing not only taking place between the five basic types of knowledge service from investigation to instruction, but also happening with knowledge within one type of knowledge service. This inherently depends on topics, learners and the relevancy for the organization. Learning objects can be composed into learning units with a generic structure that can be adapted and especially detailed with respect to the needs of topic and context of using the learning material. Parts that explain theoretic concepts need to be related with parts that provide concrete examples and case studies as well as links to experts, examples or products of the organization. Moreover, learning objects should enable learners to test their knowledge by providing questions for self-evaluation, exercises and case studies. To sum up, a learning unit can be structured as follows: introduction including motivation, relevance and challenges, contents, i.e. approaches, concepts, methods, theories, empirical results, tools, case examples, implementations, prototypes, case studies, experts and institutions, market overview, questions for self-evaluation, exercises, sources like references to literature, links to other sources, URLs and current projects, scientific or commercial studies. Example: learning units in Open Text Livelink. An easy and straightforward way to implement a learning unit is to set up appropriate structures and rules within a document-oriented KMS, e.g., in Open Text Livelink. The basic structure is implemented by creating folders and subfolders (Figure 4-23). Permissions are assigned on learning unit level and thus detailed rights to access, add or modify contents or structures can be assigned to groups of learners, tutors and teachers. Every unit can be described by customizable meta-data that allows to link it to expertise levels, skills or other topics. Pre-defined search queries, e.g., “show all elements containing a general introduction”, “show all case studies”, “what’s new this week” allow to access the objects independently of the storage structure. The learning unit may contain links to other knowledge elements of the system not specifically designed for learning, e.g., documentations of products, scientific papers, guidelines and forms. Meta-data and content can be exported in XML markup and transformed with XSLT (section 3.1.5, 165ff) to generate alternate views or maps of the learning object. The learning unit can be

343 Life-cycle

Composition into learning units

344

4. Knowledge Services

enriched with collaborative functions like recommendations of objects to other users, discussions, task lists, etc. A full-text search engine allows to search the content based on keywords and meta-data.

Figure 4-23. Screenshot: Learning object implemented in Livelink

Altogether, learning units might be a promising way as their implementation in Livelink is easy and straightforward. However, the system does not support functions specialized on learning like definition of dynamic learning paths, learner models or examinations.

Definition of learning goals

Learning goals and strategy. Before learning contents can be designed, some preliminary steps that deal with general planning of the presentation of learning contents need to be undertaken. Learning goals need to be defined that describe knowledge that learners should possess or behavior learners should show after a successful learning process. Learning goals are widely discussed in literature. Critics note that if they are open to the learner, they may be constricting because the learner focuses too much on the learning goals when working through the material. Nevertheless, they enable learners to form expectations and give

4.5. Instruction Services

feedback with regard to the success of learning. Learning goals are a foundation for creating examinations and tests and last but not least, an important instrument to plan learning units or courses. Learning strategy determines which role the learner has. Two paradigms can be distinguished: instructional paradigm and problem-solving paradigm. Design following the instructional paradigm represents a behaviorist view. The task of learners is to absorb contents chosen by a tutor or by the learning system. After completion of a learning unit, learners then get positive feedback to be motivated to continue. In contrast, the problem-solving paradigm takes a constructivist view and regards learning as an active and dynamic process. Learners autonomously work through the content and relate it to their cognitive structure. The learning environment needs to be flexible and motivating for the learner to acquire contents, e.g., in the sequence he prefers most. Special content types. In EKIs, in addition to more traditional content types like text or figures, other content types like audio, animation and video are regarded to be well-suited to design and enrich learning contents. Three general types of audio can be distinguished: speech, sound effects and music. Speech can be applied to mediate information, to substitute text or to control attention (e.g., by explaining a picture). Music and sound effects are well-suited to stimulate emotions and can have the functions to constitute a setting (e.g., digeridoo sounds that accompany an Australian landscape) or a certain period of time, e.g., music of the 1980s, identify recurring persons or situations, e.g., by repeating themes, to prepare the learner for changes in the presentation, e.g., by effects that mark the transition to the next sequence, or to control speed of a presentation. Animations are widely used, unfortunately often more to impress rather than to teach. It depends on the background of the target group and learning goals whether animations are applicable or not. They can be applied to decorate the learning environment, but may distract learners from the actual learning contents. They can control attention or motivate the learner, e.g., by giving feedback in the form of an animation, and present things that otherwise could not be observed, e.g., dynamic processes in a chemical reactor, ecological systems, circulation of blood. In the last case, animations play an important role in building mental models. A simulation is a model of a dynamic system in the perceived reality. Traditionally, complex simulations are applied in fields where faults of unexperienced learners can lead to dangerous results, e.g., steering an airplane or operating a power plant. Simulations can be fruitfully applied in e-learning as they support explorative learning by enabling the learner to discover relationships in systems by manipulating parameters. Additionally, they allow for experimenting in areas where this otherwise would principally be impossible, e.g., simulation of breedings in biology (genet-

345

Determining the learning strategy

Audio

Animations

Simulation

346

Video

4. Knowledge Services

ics) or physical experiments in an environment with different conditions than found on earth. The use of video in e-learning usually causes high efforts, but has a number of advantages. Information density is high and the real-world scenario may be accurately reproduced. It is especially useful for illustrating complex processes and social interactions (e.g., movements in sports, sightseeing tour, visualization of rooms and buildings, product presentation, social dialogues), descriptive and authentic and thus fosters memorizing knowledge. Moreover, dynamic and visual media produce emotional reactions regarded as well-suited to gain the learner’s attention and even deepen memory. However, emotions may limit the information processing capacity of humans. Video sequences that cause strong emotions should thus be used with caution. Summing up, learning objects are targeted at (a) learning goal(s) and structured using possibly several of the content types described above. However, as learning objects are atomic units of learning, they typically have to be arranged, or composed, in order to fulfil a more comprehensive learning goal which will be described in the following.

4.5.3 Composition

Creating learning contents

Composition means arranging individual learning objects into a coherent whole targeting broader or related learning goals. This can happen manually as a pre-defined, static course consisting of a set of learning objects that are bound together by rigid learning paths. Under the umbrella term of adaptive educational hypermedia systems, much effort has been put into a more dynamic scenario. Here, composition of learning objects would be performed automatically on the basis of information about the learner, e.g., parameters stored in a learner profile such as competencies, specific interests or the focus of the planned personal development, learning environment, e.g., characteristics of the device currently in use, and learning situation, e.g., the learner needs to find a solution for a wellspecified problem as quickly as possible. Courses can be composed either with support of different classes of authoring tools or with the help of services that an e-learning platform provides. This depends on how e-learning should be implemented, e.g., either as a complex application that displaces learning on site in a class or as a system targeted at supporting only specialized tasks of the learning process like organization of courses in the context of a blended learning approach. Many vendors of e-learning platforms acquired smaller specialized companies to enrich their platform with authoring functions. Creating learning units comprises planning and structuring of courses, creating

4.5. Instruction Services

347

electronic learning units, planning resources and promoting the learning unit. In this section, we focus on composition of courses while questions of assisting deployment of courses are dealt with in section 4.5.4. Authoring tools. Creation of e-learning contents involves a rich variety of tools used to create learning units and multimedia contents. They can be divided into four classes: professional authoring, rapid authoring, lecture recording and basic tools to create content. The first class of tools aims at developing professional multimedia elearning courses that allow for complex interactions. They provide extensive functions and high degrees of freedom to create different courses. Authors need to be experienced content developers, usually with programming skills. Examples are Adobe Authorware, SumTotal Toolbook and Trivantis Lectora. To support easier and faster creation of learning contents, rapid content development tools provide more standardized and pre-structured ways to create, use and manage learning units. Authors do not need to have equally advanced technical skills. However, these systems provide less flexible ways for creating desired results. Examples are ISM EasyGenerator, Adobe Acrobat Connect or Dynamic Media Dynamic PowerTrainer37 A subset of rapid authoring tools specifically targets a blended learning concept and allows for recording and enriching presentations. Corresponding tools record images of the “desktop” presented to an audience, sometimes called “screen grabbing”, e.g., a Powerpoint presentation which is manually annotated by the presenter. Typically, additional media can be recorded, such as audio and video of the presenter which are synchronized with the desktop recordings into multimedia units in real-time or after finishing recording. Finally, these units can be enriched, e.g., by structuring the contents into chapters that can be directly accessed. Examples are Camtasia Studio or IM-C Lecturnity. The fourth category comprises a variety of tools applied to create (multimedia) content. These are software products to create audio and video content, programming tools with a focus on programming languages predominant in the Web area like Java, JavaScript or ASP as well as editors to create HTML pages. Examples are Adobe Flash for animations, the integrated development environment Eclipse for programming environments as well as Microsoft Expression Web for the class of HTML editors. Principles for designing multimedia contents. Developers of courses need to consider some design guidelines and patterns that aim at fostering learning. As an example, we will present principles based on the SOI

37

Dynamic Media was acquired by IM-C in June 2008. Products thus might be renamed or integrated with learning management solutions provided by IM-C.

Professional authoring

Rapid authoring

Lecture recording

Basic tools

348

Assumptions of the SOI model for multimedia learning

Multimedia principle

Contiguity principle

Modality principle

Redundancy principle Coherence principle

4. Knowledge Services

model (Selection, Organization, Integration) for multimedia learning (Mayer 2001). The SOI model builds on three assumptions about human cognition: (1) Humans possess two different channels for receiving information, one for visual, non-verbal information and another one for auditive, verbal information. Changing between these channels demands high cognitive resources and thus switching between contents that address different channels should be applied cautiously. (2) The information processing capability is limited in that information to be processed must be held in a working memory that can only store up to seven (+/- two) so-called chunks of information. Thus, learners need to repeatedly go through e.g., text, figures or animations until they have fully acquired the contents. (3) To memorize a subject sustainably, humans need to construct a coherent mental model of it. This happens by a number of active cognitive processes of inference, comparison, generalization, enumeration and classification. Six general design guidelines can be distinguished for creating multimedia content: The multimedia principle encourages not only to enrich and decorate text with graphics, but rather to meaningfully relate text and graphics by visualizing its organization and structure or clarifying relations between things, e.g., with the help of diagrams. Ideally, learners should be motivated to actively deal with the subject.38 The contiguity principle refers to the arrangement of text and graphics. Searching and connecting text and graphics needs additional cognitive resources and thus it must be secured that text and graphics should always be simultaneously visible on screen, that words and explanations are located closely to corresponding graphics. Pop-up texts that appear when the mouse pointer is located over an item are an example of how to implement the contiguity principle. The modality principle refers to partitioning into auditive and visual channel for information perception. If the load on one of these channels is too high, the learner cannot process the whole information presented. By presenting information as speech, the visual channel is disburdened. This is especially useful if a figure or animation needs to be explained. The redundancy principle states that the same content should not be presented simultaneously as written text and speech as both channels may affect each other negatively in this case. To motivate learners, e-learning content is often enriched by background music, entertaining elements like animations or stories that are not directly related to the subject. However, results of research on cognition argue for a reduction of contents not directly related to the subject of learn-

38

Rich examples for the power of visualization can be found at URL: http://www.visual-literacy.org/, last access at 2008-10-13.

4.5. Instruction Services

ing. Reasons are that learner’s attention is attracted to unimportant information. The learning process is interrupted and information that is actually unimportant may be used to internally structure and organize the subject. Personalization principle calls for a personal language style that directly addresses learners. This stands in contrast to the wide-spread view that an objective style of writing is most appropriate. It is assumed that the computer to a certain degree acts as a partner for social interaction. Agents that guide the learner through the contents can help to mediate more closeness between learner and the learning material. Nevertheless, the appropriateness of agents and personal style largely depends on the target group. Table 4-13 summarizes six general principles for designing multimedia. Table 4-13.

349

Personalization principle

Six multimedia design principles (Clark/Mayer 2002)

principle

description

1. multimedia principle

use words and graphics rather than words alone

2. contiguity principle

place corresponding text and graphics near each other

3. modality principle

present words as audio narration rather than onscreen text

4. redundancy principle

presenting words in both text and audio narration can hurt learning

5. coherence principle

adding overly attention-catching material can hurt learning

6. personalization principle

use conversational style and virtual coaches

Structuring contents. While multimedia design principles target the presentation of multimedia contents in general, the following hints on structuring are specific to the composition of courses out of individual learning contents or objects. Segmenting denotes dividing the whole body of learning contents into single learning units. Sequencing stands for determining the order in which learning units are to be presented. This is not an easy task as there usually are multiple alternatives to divide and sequence contents, e.g., foundations can be presented separately at the beginning or “on demand” not until they are necessary for understanding. The “logic” of the contents itself usually is not sufficient to give an orientation. When presenting two or more different subjects, two basic types of sequencing can be distinguished: linear and spiral sequencing. Linear sequencing discusses a subject until the desired competency level is reached and then changes to the next. Advantages are that learners can concentrate on one subject and materials can be organized more easily. Spiral sequencing presents every topic in multiple cycles. At first, the foundations of the different subjects

Segmenting and sequencing contents

350

4. Knowledge Services

are presented. Then, all sub-topics for all subjects are detailed successively (Figure 4-24). linear sequencing

spiral sequencing

topic

topic

topic

topic

topic

topic

A

B

C

A

B

C

Figure 4-24. Linear and spiral sequencing Three generic phases

A basic scheme for the structure of single learning units is their separation into three basic phases which is applicable for both learning strategies presented above. (1) In the preparation phase, learners get an overview of the course and its contents. Existing knowledge is activated and related to the new knowledge to be learned. Examples for activities during this phase are explanation of learning goals, motivating and emotionally involving users or participants, testing existing knowledge through a short examination and kick-off workshops. (2) Acquiring is the next phase. Here, learning contents are presented or learners individually work through them. (3) The last phase, reinforcement, has the goal to secure results and check how much progress has been made. This can happen by means of small interspersed questions, recapitulation of contents, giving hints for further materials (e.g., links, literature) or offering opportunities for communication. E-learning standards. There are numerous standards to support exchanging, reusing and integrating learning contents and differ with respect to extensiveness, complexity and implementation into existing platforms and tools. Predominant aim is to secure interoperability between learning systems. Several organizations, e.g., Aviation Industry Computer-based training Committee (AICC), the IMS Global Learning Consortium (IMS/GLC) or IEEE Learning Technology Standards Committee (LTSC) have developed meta-data schemes with a focus on e-learning. Additionally, several organizations such as CanCore or CELTS derive their own proposals from established meta-data schemata. Some companies like CISCO can also be added to the large and intransparent group of initiatives. However, IEEE Learning Object Metadata (LOM) seems to be the most prominent exponent in the bulk of standards and specifications for describing learning

4.5. Instruction Services

contents. It is the master meta-data standard which provides the model for other meta-data standards provided by institutions catering for the e-learning community. An example is IMS Learning Resource Meta-data Specification which has been (re-)modeled on the LOM standard. We will briefly discuss the most wide-spread specifications. The LOM (learning object meta-data) standard specifies meta-data for certain types of digital or non-digital learning objects39 that can be used, re-used or referenced during any learning activity such as teaching programs, time schedules, learning objectives, slides, lecture notes and Web pages. Meta-data includes educational or pedagogical features like pedagogical type (e.g., active or expositive), courseware genre (e.g., hypertext or exercise), approach (e.g., inductive, deductive, explanatory), granularity (e.g., course, unit, lesson) as well as data about the educational use of the resource like roles applying the resource, prerequisites, educational objectives, target audience, difficulty and estimated time to work through the contents. Moreover, the meta-data descriptions contain data about the lifecycle of the resource, technical features of the resource like format, information for rights management and relation to other resources. The LOM standard is increasingly implemented in current learning platforms and was defined by the LTSC (Learning Technology Standards Committee) of the ISO IEEE (ISO Institute of Electrical and Electronics Engineers). Educational modeling language (EML) is a comprehensive notational system to describe units of study e.g., courses, course components and study programs. It consists of learning model, i.e. learning context and educational approach, unit of study model, e.g., learner characteristics, learning objectives, domain model, i.e. characteristics of the subject, and theories of learning and instruction, e.g., information about perspective on learning such as behaviorist or cognitivist. EML is the most powerful, but also the most complex standard and thus more demanding for authors, tutors and programmers which could hinder its wide-spread use. PAPI (public and private information) of the IEEE LTSC is a specification of a learner model, i.e. for data that characterizes a learner and her knowledge or abilities. The model includes personal data, data about relations to other users, security credentials, learning preferences, performance-orientation, and portfolio information, i.e. a collection of a learner’s works that illustrate his abilities. CMI (computer managed instruction) is a standard originally defined by AICC and targets interoperability of CMI systems. It includes extensive data enabling integration of trainings into learning management systems. 39

These learning objects should not be confused with the learning objects described above.

351

LOM (learning object metadata)

Educational modeling language

Public and private information

Computer managed instruction systems

352

4. Knowledge Services

SCORM

The shareable content object reference model (SCORM) is a collection of specifications aiming at interoperability, accessibility and reusability of Web-based learning content. It is a recommendation of the advanced distributed learning initiative (ADL) and is used as a reference model to define requirements for, e.g, the LOM specification. SCORM-compliant learning content can be delivered by any SCORM-compliant learning management system. Thus, SCORM consists of two specifications. The content aggregation model specifies how learning content needs to be delivered so that it can be run by any learning management system. A SCORM package consists of one or more assets, such as text, image or Flash movie. SCORM information is handed on in the form of an XML document instance as part of the folder or (zip) file holding the package. It includes specifications of the files making up modules and the structure of modules making up the entire package of learning content as well as metadata describing the content. The run-time environment specifies an API for communication between content and learning management system once it has been started so that it can e.g., report test results, track time spent on learning content or store the point reached when finishing a content, called last location, so that the presentation can resume at that point when later restarting the package. The IMS/GLC has provided a large number of related standards in the e-learning domain over the years since it came into existence in 1997 which aim at easing the integration of e-learning-related (meta-) data across e-learning tools, platforms and systems: pedagogical approaches: IMS Learning Design Specification provides a language used to express a wide range of pedagogies in e-learning that can be deployed by a run-time environment able to present learning material adhering to a variety of different pedagogical approaches, composition of learning material: For example, the IMS Content Packaging Specification or the IMS Simple Sequencing Specification specifically target the composition of learning material into larger units, annotation with meta-data: Other standards target the specification of meta-data to learning units modeled on the LOM standard, the IMS Learning Resource Meta-data Specification, or to learners, the IMS Learner Information Package Specification, assessment: the IMS Question and Test Interoperability Specification standardizes the definition of items used for testing and evaluation as well as for reporting the results of such assessment activities, portfolio: the IMS ePortfolio Specification focuses on the personal management of learning portfolios by learners independent of the providing institutions or learning management systems.

SCORM content aggregation model

SCORM runtime environment

IMS Specifications

4.5. Instruction Services

353

4.5.4 Consumption Composition services conclude design efforts for learning resources in general and courses with didactically sensible learning paths taking into account one or more learner types and learning paradigms in particular. The relationship between composition and consumption can be best described as deployment, the connecting concept between design time and run time. Deployment marks the transition of system components, or services, into their productive run-time environment. In complex software systems such as an EKI with loosely coupled, interacting services, deployment is a major task that comprises e.g., releasing, activating, adapting or updating existing services and contents. While this is a general issue in EKIs, the focus here is to integrate complex courses into learning (content) management systems that also provide other services related to the simple completion of courses. The term consumption is used here to comprise services that support a learner in using the designed learning material. The term consumption should not mislead, though. By no means, this is a passive process, but involves active learner participation in order to perform pre-defined lessons, or courses, select navigation paths or being aware of and interact with other learners engaged in the same or related learning processes. Figure 4-25 gives an overview of the most important steps that are to be supported by consumption services which are discussed in the following.

Deployment

Concept of consumption

behavior

complete steps

deliver

record coach

learning content elements multimedia stream

learner information parameters

ratings

Figure 4-25. Consumption procedure (after IEEE 2002, 15)

The first step in the consumption procedure is typically supported by a learning management system (LMS) that transforms learning contents into a stream of multimedia data that is presented to the learners. In order to do so, the LMS might simply replay a composed course, e.g., in the form of a Web-based training unit. It might access its own database containing learning contents, or it might refer to other repositories for retrieving learning contents, e.g., learning object repositories that are then sequenced dynamically which conceptually would mean dynamic composition, i.e. arranging learning contents during run time.

Deliver

354

4. Knowledge Services

Study

The term study is used here to reflect working with the learning contents delivered to the learner, i.e. the learner uses the courses in the more or less pre-defined way. Study also reflects the objective that instruction services present pre-packaged learning units that are not only accessed and briefly sifted through, such as many contents are that participants study in self-guided learning processes supported by investigation, individuation, interaction and also in-form-ation services (sections 4.1-4.4). On the contrary, the courses are intended to be studied intensively from beginning to end until the composed course is finished and success or failure of the learning activity can be evaluated with the help of assessment services (section 4.5.5). Generally, coaching refers to a systematic, usually longer-lasting relationship between a trained professional and an individual or team with the aim to guide development of (professional) competencies and remove barriers so that an individual or team can achieve wanted, sustainable changes in their (professional) lives. With respect to instruction services, the coaching step is much more concrete and aims at helping learners to query and select repositories holding learning content for learning experiences, navigate it along selected learning paths and to guide adaptation and delivery of learning material fitting the learners’ needs. The coaching step is highly connected to the other steps in the consumption procedure. It uses information from several sources, such as general user information available in the EKI, information gathered from the learners themselves, e.g., a standardized learner profile, e.g., PAPI (section 4.5.3), evaluation, i.e. assessment information section 4.5.5), the learner records data store (see step recording below), and the repositories holding learning resources. The final step is to record data about the learners’ behavior, i.e. information about the interaction of the learners and the courses they complete which have been selected, prepared and delivered with the help of the coaching step. A database of learner data may store data about the past, e.g., historical learner records, performance, past preferences and decisions about navigation paths or types of media preferred, the present, e.g., the current status of work with a course, suspended courses, current assessments for suspending and resuming sessions, the future, e.g., learner, or employer objectives that might have been elaborated with the help of a (real) coach. In the following, a number of tools and systems are briefly reflected that are already available for supporting parts or all of the steps identified as belonging to the procedure of consumption: Learning management systems emphasize the administration of courses, learners and course materials of the learning process rather than creation and use of learning contents, e.g., by providing calendars, news and workspaces for distribution of electronic documents for courses held on-site at e.g., a (corporate) university.

Coach

Record

4.5. Instruction Services

355

Collaboration tools support collaboration between teachers and learners and among learners. Their functions are discussed in section 4.3. Platforms for synchronous learning support simultaneous communication supported by whiteboards, application sharing, ad hoc-questions, co-browsing and file sharing (see section 4.3.4). Web-based platforms are closely related to WCMS. In fact, they can be regarded as WCMS specialized for the learning domain and sometimes are called learning content management systems (see section 4.3.5). We define e-learning platforms broadly and as being characterized by five functional areas. In the following, these areas guide the discussion of functions provided by an e-learning platform. E-learning platforms provide an integrated environment for organizationwide management of learning modules and for computer-supported cooperative learning with services for (1) presenting and using learning contents, (2) for communication and collaboration between participants and (3) to manage learner data, courses, contents and learning processes.

Definition of elearning platform

(1) Many e-learning platforms or suites are based on Web technologies. Participants access the system with a standard Web browser. Offline learning without a direct connection to the learning system requires special software clients to synchronize contents as well as other data like results of learner evaluations or profiles gained by observing the learners’ actions. Teaching and learning usually happens with the help of virtual classrooms. Additionally, learners have individual workspaces to maintain their personal information like bookmarks, contact lists, course entries and documents. Interfaces to other systems allow to access and import data like news, literature references and contact information. (2) Communication functions allow for synchronous or asynchronous communication between participants like email, blackboards, whiteboards, chat, audio or video conferences (section 4.3). Advanced functions aim at relating communication to the contents presented, e.g., to jointly solve given tasks and create an appropriate solution. Awareness functions give feedback about actions and status of other participants, e.g., degree of activity in a blackboard, users currently online, changed contents, etc. Communication functions should be regarded as an important way to spark involvement and activities of learners. (3) Administration usually happens decentrally for tasks like maintaining the contents of courses or defining access privileges for special groups and/or roles of users. Other tasks are usually centralized, e.g., technical tasks like systems maintenance, backup and maintaining connections to other systems. Learning platforms can provide programming interfaces to extend and customize their functionality. This often is deployed by specialized consulting companies or consultants of the product vendor.

Presenting and using contents

Communication

Administration

356

4. Knowledge Services

Example: Stud.IP at the University of Halle-Wittenberg. Stud.IP is an open source learning management system that provides extensive functions for students, scholars and administrative staff to support activities around management of as well as communication and collaboration around on-site university courses. Examples for functions are discussions, schedules, document folders, chat, wiki-Webs and news to support courses, interfaces to other systems like library catalogues, personal information management functions like address book, calendar, individual schedules as well as managing courses and resources with the help of an integrated room management, event timetables and staff directories. Figure 4-26 shows the personalized MyCourses page of Stud.IP which provides a good overview of the current state of all courses that the teacher or student is assigned to.

Figure 4-26. Stud.IP at the Martin-Luther-University of Halle-Wittenberg

The system was introduced in winter term 2003 at the School of Business and Economics and in February 2004 throughout the entire University. Integration with other initiatives aiming at creating ICT support for administrative tasks in the areas of university-wide planning and informa-

4.5. Instruction Services

357

tion management, e.g., personnel planning, management of student records and grades, was seen as a success factor. Since only a standard Web browser is needed to access the system, efforts of technical introduction were limited to installing the system on a centralized server. Half a year after introduction, already 848 courses of 320 institutes were hosted (October 2004). This figure has soared to a total of 19,511 courses including all courses offered by Martin-Luther-University of Halle-Wittenberg and courses from previous semesters and over 25,000 users (November 2008). During the semester, about a quarter of all registered students use the system regularly once per working day. It is used for room reservations and as a foundation for the creation of the online and published version of the directory and timetable of lectures. Currently, teachers apply it for news and timetables, for enrollment, for distributing electronic documents and for communication between teachers and students in discussion forms and over the Stud.IP mail system. Students organize their studies with the help of the system and use it for communication with peers as well as for uploading student reports.

4.5.5 Assessment In addition to enriching, composing, delivering and assisting in navigating learning contents, the learner or other stakeholders can get feedback about the effects or outcome of the learning activities with respect to the defined learning goals. This is where assessment services are needed. They allow for designing and deploying tests, quizzes and questions as well as reporting results back to stakeholders or systems. Figure 4-27 gives an overview of the steps necessary to complete an examination as part of an e-learning initiative. The individual steps are described in the following. steps

elements

create items

construct test

attempt

process response

feedback

questions

assessment

session

response

integrated feedback

scores

selected items

interaction

outcome

test feedback

Figure 4-27. Assessment procedure

The first step in assessing learners comprises the creation of questions and answers as well as the distinction between correct and false answers denoted as scores. Questions, answers and scores together are called items. There is a specification by the IMS/GLC on the design of test items. The following item types are listed in the specification:

Create items

358

Construct test

4. Knowledge Services

true/false questions: represents a simple choice between exactly two alternatives, multiple choice questions: presents a set of choices to the candidate. The candidate's task is to select one or more of the choices, up to a defined maximum number, multiple response: a container for multiple values all drawn from the value set defined in the item, image hot spot: graphic interactions involving images with specially defined areas, select text: presents a set of choices to the candidate represented as selectable runs of text embedded within a surrounding context, fill in the blank: contains a number of gaps that the candidate can fill from an associated set of choices, e.g., free-text fields for short or extended statements to be inserted or images to be inserted into a gap by the candidate, slider: presents the candidate with a control for selecting a numerical value between a lower and upper bound, drawing: allows the candidate to use a common set of drawing tools to modify a given graphical image (the canvas), order objects: the candidate's task is to reorder the choices, the order in which the choices are displayed initially is significant, match items: presents candidates with two sets of choices and allows them to create associates between pairs of choices in the two sets, but not between pairs of choices in the same set, associate items: presents candidates with a number of choices and allows them to drag objects or drag targets and thus create associations between them. As all IMS specifications, this list is not designed with a particular pedagogical approach in mind. The specification supports interoperability between e.g., authoring tools, banks of tests and other assessment items, test construction tools, learning management systems or assessment delivery systems. If needed, it can be extended with proprietary assessment items by the designers of evaluation services. Authors of items are professionals, experts or teachers. They are supported by item bank administrators who manage the pool of items created in the past. In this specification, assessments are structured into sections which in turn consist of items, i.e. the individual questions. A test constructor develops a test for a specific target group using the available test items from the item bank. This procedure can be semi-automatized, e.g., by automatic selection of test items and by so-called “scrambling” which is changing the order of answers to multiple-choice questions. The output is a test, called assessment consisting of an ordered list of selected items.

4.5. Instruction Services

The assessment is then presented to the candidate who will step-by-step interact with the individual test items. The participant performs the test which can be observed by a supervisor. Interactions between candidates and assessment are called sessions. An attempt consists of one or more sessions on the same test resulting in responses to the questions which are recorded by the assessment tool or system. The recorded results of the interactions by the test participants are then rated in order to determine correct and false answers and aggregate the responses into test outcomes. The final step of the assessment procedure presents the outcomes to the candidate or other stakeholders. Feedback can be presented integrated, i.e. as part of the test, integrated into the test items, modal, i.e. on its own, separated from the test items, test, i.e. after processing all responses, based on the values of the test outcomes. Evaluation of learners might aim at immediate feedback about their learning progress. Most straightforward ways for implementation is direct evaluation of learners, e.g., after they worked through a learning unit. Descriptive feedback and showing learners ways how they could improve their skills should be preferred in place of short feedback that just enumerates failures. Naturally, this is far more complex to implement. Examinations go one step further by incorporating the goal to certify learners. Online examinations head for a computer-based realization of examinations with the goal of making the evaluation process more efficient. Standardized assessments can result in issuing trusted statements that certify that a knowledge worker has participated successfully in an assessment and thus supposedly holds a certain level of competence with respect to a defined knowledge domain. The standardized format of test items and reporting of results paves the way towards an integration of assessments into the EKI and reuse it when offering other knowledge services. More advanced uses of evaluation might include complex internal models of the learner that are used e.g., for investigation: filtering and presenting results of discovery services according to the presumed needs of the knowledge worker, individuation: updating the profile of the learner with test certificates which have been successfully achieved, interaction: suggesting peers, (slightly) more advanced co-workers or experts in the domains that the learner has shown interest in according to the current level of competence, in-form-ation: providing help with formalizing contents that is tuned to the competence level of the knowledge worker and, last but not least, instruction: personalization of learning contents, i.e. suggestions for contents to be presented and the next steps on the learning path.

359 Attempt

Rate response

Feedback

Instant feedback

Certification

Integration in EKI

360

4. Knowledge Services

Case Example Mike works as programmer in the IT department of a medium-sized automotive supplier. He has a professional IT education and afterwards received a Bachelor degree in business information systems. Mike is very fond of Web 2.0 technologies and does not understand why his company does not benefit from them. The intranet consists mainly of a Groupware server, an ECM and a unified communication server (UCS) that supports instant messaging, voice telephony and desktop video conferencing. All three systems are sophisticated products within their product group, but lack integration and additional modules to form an EKI. Mike misses, e.g., a possibility to show the project team calendars managed in the ECM within his personal calendar, stored in the Groupware system. He would also like to store email attachments directly within a team room of the ECM, but the Groupware client does not support this. Finally, he would like to see the presence information, available in his UCS directly within the ECM and Groupware server, so that he could decide easily whether to answer an email with a phone call or better with an email reply. Mike uses an external social networking website to keep in touch with his friends from school and university. He would like to see such functionality within his company as well, because programmers are spread across multiple departments and it is not easy to keep in touch or identify the best contact for a given problem. He is also actively engaged in writing a weblog to share his ideas about new technologies, but he cannot blog as often as he would like to, because some topics are not to be shared with external people by company policy. However, there is no internal blogging software, as it is considered a waste of time to write a personal diary. To keep himself informed about the latest trends in software development and computer industry, he has subscribed to a number of newsletters and additionally is screening RSS feeds from a few technology websites for news every day. Unfortunately, he has no idea how to best store interesting news or articles. For some information pieces he uses a browser plug-in to file them directly to disk as HTML. Some articles can be downloaded as PDF or can be easily printed to PDF and stored as files. Finally, he also uses a social bookmarking service to keep track of interesting news. To make his colleagues aware of them, he occasionally writes emails with links or file attachments, in other cases he posts links and PDFs on team rooms. All in all, it is difficult for him to retrieve knowledge elements later on. To make things even more complicated, one of Mike’s colleagues set-up an unofficial Wiki system that a few programmers now use to document technical know-how about their programs. Mike’s current project is to develop an add-on for the ECM in order to manage learning objects. Colleagues from an IS department have evaluated existing LMS, but found nothing with integration into the company

4.5. Instruction Services

infrastructure, so that it was decided to build one on their own. Since collaboration and communication functions are already available, the LMS only needs to care about managing courses and associated learning objects as well as keeping track of learning success and competencies in learner profiles. Although the task seems not very complicated in the beginning, Mike and his team found out very soon, that learner profiles might be harder to implement than expected since there is no central user directory that holds all required information. Therefore, they would have to extract information from every system and integrate them in their learner profile or otherwise profiles would require a lot of manual maintenance.

361

362

4. Knowledge Services

Questions and Exercises 1. What are differences and similarities between search for Internet

resources and for resources in an organization’s Intranet? 2. Find examples for cases in which on the one hand search by keywords

and on the other hand by navigation and browsing visualizations of resource collections is most appropriate. 3. Draw an example for a knowledge map that shows which types of knowledge and resources are needed during the phases design of questionnaire, data collection, data analysis and creation of final report of a marketing research company. Give concrete examples how this map should be generated and applied. 4. A medium-sized company in the automotive industry wants to introduce a DMS. Typical types of documents are application for leave, construction plan, internal mailing to all employees, invoice, e.g., for office materials, and long-term contract with customers. Analyze which document types should be managed within the DMS. Therefore, explicitly name decision criteria and use them for classification of documents. Decide which documents should be held online, nearline and offline. 5. What is the difference between non-coded information (NCI) and coded information (CI) and what are advantages and disadvantages of each type for archiving documents? 6. Compare Groupware, DM and e-learning systems with regard to (a) goals, (b) focus on supported tasks, (c) role of users and (d) characteristics of system contents. Draw a conclusion on main similarities and differences of these system classes. Classify managed contents according to the dimensions of knowledge presented in section 1.1, 3ff. 7. What are the similarities and differences of the terms information resource, content, asset and document? Please give concrete examples for each term. 8. Define the term expertise and the stages of expertise distinguished. 9. Describe the historical development of WCMS, discuss the supposed benefits of additions in each phase and give an outlook on further developments. 10.Describe a concrete example of a communication process structured according to the Shannon/Weaver model. 11.Draw a conversation network that structures the following conversation at the desk of a library: A student comes to the University library and says: “Hello, I would like to extend the due date of these books.” The clerk at the library desk might answer “Sorry, but this is not possible. These books are ordered by another person.” or “No problem, I have postponed the due date for two weeks.” If the books have already ben

4.5. Instruction Services

ordered by someone, the student could try to get at least two more days. If that is also no option, then the library clerk can at least offer an extension of 24 hours. If there was no extension possible at all, the student returns the books. In either case, he says goodbye before he leaves. 12.Give a concrete example how ICT can support a conversation that was structured with a conversation network. Which classes of systems are most suited to support structured conversations? 13.List and explain functions of an e-learning platform required to support University courses following (a) a blended learning and (b) a distance learning approach. 14.Characterize how e-learning modules should be designed from the perspectives on learning of behaviorism, cognitivism and constructivism. 15.Explain principles of designing learning units and multimedia content. Provide a checklist for software developers which should guide them in actually designing contents. 16.Give an overview of the most important e-learning standards. 17.Describe how learning objects are structured, can be supported and discuss their potentials with respect to reuse. 18.Define e-learning platforms and describe their main functions. 19.Make at least three proposals for extensions that substantially enhance the learning (content) management system used at your university.

363

364

4. Knowledge Services

Further Reading Asprey, Len; Middleton (2003): Integrative Document and Content Management: Strategies for Exploiting Enterprise Knowledge. Idea Group Publishing, Hershey The first chapter of this book outlays the architecture of an integrated document and content management (IDCM) and gives a brief historical overview over the historical development of document and content management systems. The features of the IDCM are examined briefly in the following chapters. Baeza-Yates, R.; Ribeiro-Neto, B. (1999): Modern Information Retrieval. Addison Wesley, Harlow This book comprises in-depth explanations and comparisons of concepts and techniques for all important phases of information retrieval, i.e. preparation of resources, indexing, query operations, searching and presentation of results. Additionally, they illustrate how searching in special content types (multimedia IR) and in special domains (digital libraries) can be realized. Boiko, Bob (2005): Content Management Bible. 2nd edition, Hungry Minds, New York This book presents a comprehensive overview over content management and should be seen as a practical guide for doing content management with a focus on Web content management. On over 1000 pages content, content management, content management projects, the design of CMS and building as well as implementing a CMS are discussed. Borghoff, U.M.; Schlichter, J.H. (2000): Computer-Supported Cooperative Work: Introduction to Distributed Applications. Springer, Berlin This book includes all fundamentals about collaboration services. It extensively explains basic principles and concepts of Computer-Supported Cooperative Work (CSCW) like history and goals or concepts of asynchronous and synchronous communication as well as applications to support CSCW like communication systems, shared information spaces, workflows, workgroup systems and multiagent systems. Carliner, S., Shank, P. (eds.) (2008): The E-Learning Handbook: Past Promises, Present Challenges. Wiley, San Francisco This is one of a large number of contemporary and comprehensive accounts of learning technology which also touches the relationships with other types of knowledge services such as interaction services and Web 2.0-induced changes to the needs for support of learning processes.

5 Access Services

The knowledge services discussed in the previous chapter should not be isolated from other applications a knowledge worker uses. On the one hand, users want access to them from within their usual work environment and they want access any time and any place. Therefore, mobile access and desktop-integrated access play an important role for utilization and user acceptance of EKIs. On the other hand, there are some demands imposed by the need to run these infrastructures cost-effectively. This requires solutions that run on any desktop, independently of the operating system and the installed applications. In addition to that, control over installed versions and software use is required to manage updates and provide efficient user support. These two demands lead to a variety of alternatives that are either quite well established like browser-based application access or just emerging such as rich thin client technologies. Ultimately, users want a single point of access to all services offered within the EKI, so that they are easy to use and can conveniently be accessed using single sign-on. From a cost and administration perspective, central user management and profiling capabilities are beneficial (see ”Person dimension” on page 191). Access technologies are classified into server-based technologies, e.g., portals and server-side personalization technologies which are presented in section 5.1. Client-side access technologies can further be divided into thin clients, rich thin clients, rich clients as well as desktop integrated access and are presented in section 5.2. Finally, section 5.3 deals with the basic requirement of mobility which is typical for knowledge work (section 1.2, 21ff). After studying mobile computing in general, the specifics of mobile devices and mobile application are discussed with respect to their differences to more traditional access to EKI from desktops.

Overview

On completion of this chapter, you should be able to define server-side, client-side and mobile access technologies, analyze potentials of each of those classes for accessing EKIs, define user model and use important characteristics for analysis of user models managed in various software solutions, employ the concept of personalization to tailor knowledge services to the needs of participants, define portals and analyze their main functions and application areas, assess the overall contribution of access services to usage and user satisfaction when implementing EKIs.

Learning objectives

366

5. Access Services

5.1 Server-side Access Technologies

Overview

Users of an EKI deal with a variety of applications, systems, services and contents. In practice, this usually leads to challenges since they continuously need to switch between multiple systems that regularly require logon with dedicated user names and passwords, have different interfaces and different ways to select and present data, information and knowledge. In this context, single point of access means integration of these various sources into one single interface with a consistent look and feel. Furthermore, there is the danger of “information overload” as access to too much information and knowledge might decrease productivity. These requirements are usually met by means of Web-based portals which we will discuss in section 5.1.1. One of the main features of portals are selection and filtering of contents according to the user’s individual needs and requirements. This is denoted as personalization and will be discussed separately in section 5.1.2.

5.1.1 Portals Portals as starting points

Internet portals

Enterprise portals

The term portal denotes a monumentally designed entry to a building. It is used as a metaphor for Web interfaces that integrate various heterogeneous information sources and applications on a single Web page. Simple, Web-based portals are thus one way to realize integration on the presentation layer. The metaphor references an important characteristic of Web portals: They are designed to provide a central entry point for users to application systems, just as doors provide to buildings. However, they do not represent a work environment, rather they are designed to provide an overview of and integrated access to various information sources and systems (represented in so-called portlets) and to guide users to those systems they need to work with, just as the hallway behind a portal leads to rooms and chambers where people live and work. The term portal is used differently in the context of the Internet and the Intranet/Extranet. Internet portals act as a major starting site for users in the Web or a site that users tend to visit as an anchor site. They provide a structured collection of links to Web sites and are often coupled with additional services, e.g., email accounts, chatrooms and calendaring. There are general-purpose portals comprising a large variety of topics and specialized niche portals focusing single topics and supporting communities in sharing information around these topics, e.g., travel, cycling, rock-climbing or knowledge management. Examples for general-purpose Internet portals are www.t-online.de and www.lycos.com. Single topic portals are e.g., www.contentmanager.de, www.kmworld.com and www.xml.com. Portals available on the Intranet or Extranet denote the actual Web site that unifies access to different applications and also a software application

5.1. Server-side Access Technologies

running on a server that provides functionality to achieve unified access. We will use the term portal in the latter sense. Portals of this type are also called enterprise portals. However, sometimes Intranet portals are understood in a broad sense that subsumes several related components including interfaces to integrate the source systems, middleware and the underlying infrastructure. This includes many services of the EKI discussed in previous chapters, especially integration services (chapter 3, 151ff) as well as application server (section 2.3.3, 137ff). However, portals are just one way to access selected services and systems of an EKI. Employees regularly will use an EKI in many different ways and settings, e.g., as extensions to their desktop, with the help of mobile devices or in shared workspaces. Put briefly, portals are designed to bundle personalized, i.e. according to user’s needs, task, project or role-specific, i.e. targeted at the actual working context of the user, location-independent, i.e. accessible with all devices able to present Web content including mobile equipment, and secure, i.e. through encryption of sensitive data as well as user authentication, access to software functions, i.e. functions offered by different systems supporting business processes for employees, partners and customers, e.g., employee self-service functionality, data, information, and knowledge, i.e. structured as well as unstructured data with context and meta-data from multiple information sources. Portals ideally provide the following functions: personalization of contents and systems, single sign-on to several systems and integration of systems on the presentation layer. Personalization denotes the ability of the portal to adapt to the specific needs and requirements of its users. This at least comprises the possibility for users to choose which elements are presented where on the Web interface and to customize the general appearance (e.g., colors, icons, decorations) of the portal. Advanced personalization includes user profiling as well as automatic adjustment and filtering of contents and functions (section 5.1.2). Users usually need to authenticate separately on every single system they use. Portals relieve users of this task by automatically transferring the data needed to systems that are accessed through the portal. In fact, this feature may be one of the major driving forces and reasons for the widespread use of portals. The necessary user-related data such as user name and password is managed and provided by account or identity management services (see ”Person dimension” on page 191).

367

Portals and EKI

Portal tasks

Functions

Personalization

Single sign-on

368

5. Access Services

Drag & relate

Portals display user interfaces of various applications in a single consistently designed user interface. Currently, there are ongoing developments for a tighter integration of application systems with portals. Most integration efforts aim at data integration, specifically to ease data transfer between different applications accessed through the portal. An example is the drag & relate technology from SAP. Users can transfer single data items between portlets by selecting them (e.g., sales order number 537), dragging them to another portlet (e.g., order portlet) and thus invoke a function of the target application that uses the data item as parameter (e.g., details of sales order 537 are displayed). SAP calls this functionality that integrates heterogeneous systems drag & relate in analogy to drag & drop on the desktop.

Target group

Content and area of application

Portlets

Classification of portals. Various classes of portals can be distinguished depending on target group, primary content and area of application. The group of users that predominantly accesses the portal can be (a) employees that use a B2E (business-to-employee) portal providing e.g., knowledge services, employee self-services or services that act as a channel for internal communication and marketing, (b) business partners such as suppliers or partners in a joint venture, that access a B2B (business-tobusiness) portal, e.g., to execute business transactions or to collaborate in joint projects with the organization hosting the portal, and (c) customers, targeted by a B2C (business-to-consumer) portal, e.g., an online shop for product sales or Web pages for advertisement and customer support. Another alternative to classify portals is according to their primary contents and areas of application into: Enterprise application portals that represent an interface to heterogeneous types of applications, e.g., enterprise resource planning systems, geographical information systems or customer relationship management systems, enterprise information or knowledge portals that supply information from various data and information sources like data warehouses, file servers, document management systems, report engines or external news feeds, as well as EKI services to handle knowledge, and enterprise collaboration portals that provide virtual spaces for task-oriented collaboration of teams spanning organizational units. Design and implementation of portals. There is a rich variety of different ways to design and implement Web pages that represent the user interface of a portal. Internet portals often try to attract users by an appealing layout. Portals used in a business context are usually designed more function-oriented, though issues such as a layout that fits and communicates corporate identity are often considered important. Figure 5-1 depicts an example of a typical portal Web site applied in organizations (based on Priebe 2004). Portlets group functions, contents or

5.1. Server-side Access Technologies

369

navigational elements similar to windows in graphical user interfaces of operating systems. Portlets display their title and functional elements (e.g., minimize portlet, print) within a headline. Below this headline, contents are displayed in the portlets’ content area. Each portlet may be an interface to a different system. Portlets encapsulate specifics of generating output for involved systems, are therefore directly connected to the systems although they run in the environment of the portal server. Many software vendors have already implemented portlet interfaces for their applications so that these applications can be easily integrated into a portal. portal content management system (CMS)

CMS portlet portlet

customer relationship CRM portlet management (CRM) system integration rendering online analytical processing (OLAP) system

OLAP portlet

...

...

decorations & controls

portlet content

portal server

Figure 5-1.

Portal server, portal and portal components

Usually, it is the task of the portlet to process and display contents of the connected system within the content area of the portlet. A portal server integrates the individual contents of each portlet, adds the portlets’ headlines together with their functional elements and produces the Web site around the portlets. It handles general interaction with users and communication between portlets (e.g., functions such as drag & relate). Product examples for portal servers are IBM WebSphere Portal, Microsoft Office Sharepoint Server and SAP Enterprise Portal. Currently, there are two standards worth mentioning in this context: The Java Portlet Specification is based on the Java 2 Platform (Enterprise Edition, J2EE) and defines the inner structure of portlets and interaction between portlet and portal server by means of Java interfaces that need to be implemented by the portlet. Naturally, the standard does not specify the contents that are rendered in the portlet. The Web Services for Remote Portlets (WSRP) standard defines Web Service interfaces for portlets that may run remotely outside the portal server. In this case, portlets can be implemented on different platforms, e.g., on J2EE or on Microsoft .NET.

Portal server

Portlet standards

370

Complexity of implementation

5. Access Services

However, Microsoft is trying to establish an own de-facto standard with their WebParts introduced with Sharepoint and now brought to ASP.NET. The implementation of a portal can easily rise to a considerable level of complexity, if several user groups and systems are involved. An example of a B2E portal of a large car manufacturer gives an impression of the complexity (Amberg et al. 2003): A typical user accesses 50 portlets per day that are connected to various systems. Some of these portlets are displayed per default, others can be individually selected and configured by users. The company has 7 business units, each composed of 15 departments that each have 10 user groups on average. This results in 1,050 user groups. The portal provides access to several thousand portlets, some providing rich functionality. This complexity is independent of the technological implementation, e.g., the type of the portal server that is used.

5.1.2 Personalization

Profiling

In a broad sense, personalization encompasses adapting formerly standardized services to specific needs of individuals or groups. This definition includes adapting the services of a computer system, customizing products or services to customer needs as well as individualized marketing. In the context of EKI, personalization denotes adapting contents and functions to users’ needs. Personalization can be based on individual characteristics, such as skill level, interests or preferred media as well as on characteristics of users’ actual work context, such as processes they are involved in, their location, time and technical requirements, e.g., work with a mobile device, and current tasks. To accomplish this, the system needs to have an internal representation of relevant data about the user, a so-called user model. For efficient personalization, the model needs to be continuously updated by profiling data about users and their actions. Figure 5-2 shows the process of profiling and the subsequent application of the collected profiles to personalize an EKI. Grey arrows visualize data flows to or from the participant. Black arrows visualize the sequence of steps taken.

participant adaptation

collection

user model analysis

Figure 5-2.

The process of profiling (based on Frielitz et al. 2002)

5.1. Server-side Access Technologies

Creating a user model can be understood as acquiring, formalizing and managing knowledge about users. The success of automated personalization depends on how accurately user models allow to determine actual user interests and information needs. Principally, a user model can be used to diagnose and evaluate knowledge and skills of a user, e.g., with an elearning system that applies learner models to adapt the type or media of contents presented (section 4.5, 331ff) based on explicit user evaluations or implicit user profiling, forecast user actions and reactions by analyzing paths of development, though this regularly is a complex task since there may be multiple interpretations of one and the same user behavior, e.g., short response times may either indicate that the user is under-challenged or that his concentration decreases, filter contents and other outputs of the system, e.g., search results, system dialogues or information presented in a portal site, improve incorrect or insufficient user input by correcting or complementing it based on supposed goals of the user, notify users and recommend topics, news and issues that may be of interest for them, assign tasks or route requests to users based on, e.g., their expert level, skills in certain domains or their current work load which may be supported by a workflow management system (section 3.4.2, 225ff). Table 5-1 gives an overview of attributes that allow for a general characterization of user models (Mertens/Höhl 1999). The subject of the user model may be either the individual user or a role, group or organization the user acts for as an intermediary. For example, in a purchasing department, characteristics and requirements of the employee’s organization may be more important than his individual preferences. Individualization may happen either on the level of individual parameters per user or by relating users to defined classes, called stereotypes. For example, collaborative filtering (section 4.3.2, 290ff) may recommend items to users based on categorization into groups that are representative for their interests. Another example is selection of contents according to the expert level of the user, e.g., novice, advanced beginner, competent, skillful master or expert (section 1.1.1, 4ff). User models can be based on different types of information. Capturing hard facts, e.g., gender, number of errors, invocations of user help, user response times, is substantially easier than collecting and storing “soft” information about the user, e.g., goals, aims, plans. Dynamic models can be changed and thus evolve during a user session while static models remain unchanged. Systems with transparent models allow users to inspect what information is stored about them, sometimes even to change it. This may be undesirable as it could bias user behavior and lead to erroneous classifications. A user model can be either valid for

371 Application of user models

Characterization of user models

372

5. Access Services

multiple sessions or only for one session. Acquisition of information about users may happen automatically by machine learning, e.g., by neural networks, or by manual addition of rules. Table 5-1.

Collection

Data privacy

Analysis

Characteristics of user models

attribute

values

subject

receiver (e.g., role) vs. user

level of individualization

individual

vs. stereotype

type of information

soft

v.s hard

dynamics

static

vs. dynamic

transparency

transparent

vs. intransparent

time of validity

multiple sessions

vs. one session

acquisition of information manual

vs. automated

collection of information

vs. implicit

explicit

Finally, the collection of information can be: explicit with the help of a number of questions that users have to or should answer, e.g., gender, age, topics of interest, area code, implicit by observing user behavior and monitoring interactions with the system, e.g., user tracking or click stream analysis, or based on a combination of data collected from other systems, e.g., enterprise resource planning systems or human resource management systems. Collection of information about users may potentially harm personal rights and usually is regulated by a number of laws. In Germany, data collection, storage, processing and use is regulated in the federal data protection act (in German: Bundesdatenschutzgesetz, BDSG), data protection acts of the federal states as well as the information and communication services act (in German: Informations- und Kommunikationsdienstegesetz, IuKDG). Article 2 of the teleservices data protection act (in German: Teledienstedatenschutzgesetz) establishes the basic principle to store, process and use no data about persons or as little data as possible. Users principally need to explicitly agree to further use of their data. To guarantee compliance with laws, data should be processed anonymously where possible, e.g., only related to roles or groups of users. Legal regulations limit application of advanced user models. Users may be more willing to accept that systems collect and use information about them, if they see benefits, provided they know what information is stored and what happens with it. Analysis of the collected information includes data mining, e.g., the selection, cleansing, transformation and analysis of relational data, e.g., skill or interest profiles, in analogy to data warehouses and customer relationship management systems,

5.1. Server-side Access Technologies

text mining, e.g., the analysis of submitted documents or of contributions in newsgroups, and web mining (or click-stream analysis) which is a technique to analyze data about user movements on a Web site, an Intranet platform or Webbased systems of an EKI. It is widely applied and supported by a number of commercial products in B2C electronic commerce. In a personalization context, Web mining is used to discover navigation paths (“foot prints”, “trails”) through Web sites, evaluate which links are preferably clicked, or how long users stay on a Web site. It may help to discover correlations between contents and layout of a site, on how they influence user actions, e.g., purchase orders after visiting certain sites, preference of links according to their position on the Web site, and time spent on a Web page. There are two basic types of adaptation: customization and personalization. Customization is under direct user control. The user explicitly selects between different options, e.g., the portlets displayed on their individual portal entry page. Personalization is driven by the system which delivers individualized pages to users based on user models. Adaptation can be accomplished with pre-defined “if-then” rules that regard data, roles, membership in groups and communities of the user as well as with event- or time-driven triggers. The problem is that designers need to anticipate users’ needs which may change over time and according to the actual usage context. Other ways are automated content-based filtering, e.g., by comparing user profiles with contents of the knowledge base, and automated collaborative filtering (section 4.3.2, 290ff), e.g., “communities of preference”, active recommendations by other users as well as automated or hidden recommendations. Examples for adaptation in EKI systems are: search results that are filtered according to users’ information needs, contents of learning units and evaluations that are selected and adapted to users’ skill levels and user preferences, recommendations of items to develop users’ skills in their areas of interest, location-based information, that are only delivered to users at the respective location, wizards that guide users through system functions according to their level of experience with the system, and dialogues that provide information appropriate to the technical skills of users, e.g., instructions, error messages. Personalization is based on the assumption that information needs of users can be identified and explicitly represented by means of a user model. This may not always be possible satisfactorily, since in many cases either needs cannot be explicated or user models become too complex.

373

Web mining

Adaptation

Critique to personalization

374

5. Access Services

Other issues are capturing and processing of sensitive user data. In some cases, users may not appreciate filtering, e.g., because they do not trust filtering methods and fear to overlook items. On the contrary, they may not recognize important items at all if contents are not filtered. Some sort of basic filtering is almost always applied, e.g., spam filtering in email systems or basic rights management in knowledge infrastructures, however, this is not individualized. Altogether, personalization is a powerful, yet complex way to support individual user needs. Its success largely depends on how much effort is taken to create appropriate models of users which need to be balanced with the benefits of personalization.

5.2 Client-side Access Technologies Access to enterprise applications in general and EKI in particular should be as easy as possible. Thus, the ideal solution would be a single software platform that is able to support the whole palette of mobile and stationary devices regardless of what operating system runs on them. Given the different capabilities of the devices, this can currently not be realized. Many applications still require the installation of a platform-dependent software client. The only help for system administrators is the use of software deployment solutions that support remote installation of software. However, these tools are mainly platform-dependent and programmers are still left with the demand to supply clients for all main platforms. In the following sections, three different approaches are presented to overcome these limitations. They attempt to make it easier for software developers to create different graphical user interfaces (GUI) from the same source code and help administrators to automate updates and centrally control program versions used throughout the company. This is required so that services offered in EKI can efficiently be accessed by a multitude of access applications and appliances.

5.2.1 Thin Clients HTML clients

Inefficiencies

A first approach uses Web browsers without any additional software as runtime environment on the client. The advantage of this approach is that applications run on every platform that offers a Web browser and that no specific client software has to be installed. However, this means that programmers can only use HTML for describing GUI elements and application logic has to reside on the server. That leads to a large number of socalled round-trips. A round-trip is the cycle from (1) sending data from a client to a server, (2) evaluating and processing data on the server and (3) sending back the

5.2. Client-side Access Technologies

refreshed user interface to the client. Consequences of many round-trips are slow applications and high network traffic. Applications are comparably slow in this inefficient approach, because server processing capacity per user is typically smaller today than client processing capacity. Network traffic is increased, because a large percentage of data transferred from server to client is identical for two consecutive round-trips. Since HTML pages for Web applications are dynamically generated on the server, they cannot be cached on clients. In addition to that, HTML was not designed to describe application user interfaces and therefore is limited in its capabilities. Script languages like JavaScript, JScript, VBScript or ECMAScript can be used to put some application logic on the client side. These script languages are mainly suited to evaluate user input for syntactical correctness, so that only syntactically correct and complete forms can be submitted to the server for processing and some round-trips are saved. In addition to that, some formatting information can be defined using cascading style sheets (CSS) so that less HTML code has to be transferred as the static CSS data can be cached on the client. Web pages that use both, script languages and CSS, are commonly denoted as dynamic HTML (DHTML) pages. DHTML overcomes most of HTML’s limitations regarding GUI rendering. For example, dynamic menus are possible as well as dynamic tree views. With Asynchronous JavaScript And XML (AJAX), the capabilities of DHTML were taken one step further. This technique allows to trigger server-side functionality in the background and deliver the results of the processing as XML data to a JavaScript callback handler, which in turn can present the results to the user as HTML that is dynamically inserted into the presently displayed page. Therefore, only part of the page is refreshed so that for users the behavior is quite similar to a rich client application. In practice, such functionality often uses comma separated strings instead of real XML data for simplicity and speed reasons, but is still called AJAX. Various AJAX Frameworks can be used to perform common functionality, instead of programming them yourself like Dojo or Yahoo! UI Library. CSS is a language that supplements HTML to define layout-oriented properties of HTML elements. The properties defined in the CSS language include text fonts, colors and sizes as well as margins, alignments and adjustments just as borders, background colors and images. They can be specified within the style attribute of an HTML element, within the head of an HTML file or within a separate CSS file that can be referenced by several HTML files. Separate properties can be specified for different output media. Examples for important media types are screen (for PC clients), handheld (for clients with small displays) and print (for data that has to be printed). Listing 5-1 shows an example of styles for the body

375

DHTML

AJAX

CSS

376

5. Access Services

of an HTML page (body), for first-level headings (h1), paragraphs (p) and list items (li) specified in the CSS language. body { background-color:#FFFFCC; margin-left:100px; } h1 { font-size:48pt; color:#FF0000; font-style:italic; border-bottom:solid thin black; } p,li { font-size:12pt; line-height:14pt; font-family:Helvetica,Arial,sans-serif; letter-spacing:0.2mm; word-spacing:0.8mm; color:blue; } Listing 5-1. DHTML drawbacks, DOM

XSLT generated XHTML

XUL, XAML

Cascading style sheet example1

However, applicability of DHTML also has its limits as different Web browsers use different ways to address HTML elements. An HTML document can be represented as hierarchical tree of elements. The scripting language can access every single element in order to read and write its properties and values. Browsers provide a programming interface called document object model (DOM) to enable that. DOM implementations are slightly different in major browsers. Therefore, developers of enterprise applications often have to create multiple versions of the client (e.g., for Microsoft Internet Explorer, Safari and Mozilla browsers). Another approach that also uses HTML on the client-side leverages XML and XSLT to generate HTML pages (section 3.1, 153ff). This technique is especially interesting if the client is capable of transforming the XML file into HTML (or XHTML to be precise which is the redefinition of HTML using XML as meta language). Client-side transformation leads to significantly reduced server and network load as only data that has to be displayed is transferred in every round-trip. However, capabilities of different browsers are again diverse, so that client-side transformation is no real option. An advantage of the XML approach with server-side transformation in contrast to HTML solutions is that different versions for different browsers or client devices can be created relatively easy by using slightly different XSLT style sheets. Recent approaches use XML-based languages to specify the GUI of an application including layout and elements. Two corresponding languages are XUL (XML User Interface Language), which is developed and supported by the Mozilla foundation and XAML (XML Application Markup Language) which is developed by Microsoft and is used within the .NET platform 3.0 and 3.5 as well as within Windows Vista. The advantage of this approach is that complex client GUIs can be built quite easily, and the

1

URL: http://www.w3schools.com/css/, last access: 2008-11-01

5.2. Client-side Access Technologies

same XML files can be used for Web-based applications as well as for rich clients. The amount of data that has to be transferred is much smaller than with traditional HTML clients, as the entire GUI description only has to be transferred once. Simple checks can be done on the client-side with scripting languages. But again, there are at least two competing industry standards so that it will not be possible to write an application that runs on both, Mozilla and Internet Explorer. On the one hand, Mozilla has a bit of a head start as XUL is working for several years, whereas XAML was introduced together with Windows Vista in late 2006. On the other hand, XAML is more comprehensive as it includes not only elements to describe GUIs, but also elements of SVG and SMIL (Synchronized Multimedia Integration Language) to describe vector graphics and animations. Java Server Faces (JSF) is a technology that corresponds to XUL and XAML, but on the server-side. JSF extend Java Server Pages (JSP) and define a set of GUI elements that are converted to HTML, before the Web page is transferred to the client. Thus, they help programmers to design GUIs, but do not ease the problems with high server and network load. Independently of the concrete technology used, thin client applications all suffer from the restriction, that they cannot run while the user is offline and despite the trend of being always online there are still numerous occasions when users are offline for some reason. This drawback should be overcome with frameworks like Adobe AIR, Google Gears, Mozilla Prism and Yahoo Browser Plus, that are quite different in technologies used, but all aim at the same target to make online applications offline available and cache data in a local embedded DBMS. All these technologies are quite young and it cannot be foreseen currently if they will be able to prevail. Offline capabilities are also addressed by the HTML5 effort that is driven by the Web Hypertext Application Technology Working Group (WHATWG) and the World Wide Web consortium (W3C) which currently is available as a draft recommendation from August 2008. It introduces new elements with some semantic meaning like for navigation areas, or for footer areas that extend the general tag. It also removes some old elements marked as depreciated in HTML4 like and that are replaced by CSS styles. It further integrates better multimedia capabilities by introducing the , and elements that allow dynamic vector graphics as well as direct playback of audio and video files without the need for a browser plug-in. Finally, it incorporates Web Forms 2.0 as a replacement of the HTML4 form element and introduces client-side caching and data storage requirements. Altogether it is targeted at facilitating better Webbased applications, richer multimedia Web sites and better structured online information. Therefore, it has a different target than the XHTML 2.0 specification which solely aims at informational pages.

377

JSF

Adobe AIR, Google Gears, Mozilla Prism, Yahoo Browser Plus

HTML 5

378

5. Access Services

5.2.2 Rich Thin Clients Plug-ins

Java applets

Java FX

ActiveX

Rich thin clients use Web browsers extended by plug-ins as runtime environments. The advantage of such an approach is that plug-ins provide a richer set of GUI elements and offer additional capabilities to implement client-side application logic. Most plug-ins are available for all major browsers, so that deployment across platforms is possible. Network load is minimized through client-side caching, so that typical operations like resorting data or updating calculated fields do not require fetching data from servers again. Functionality like drag & drop, context menus and shortcuts are also possible. Unfortunately, plug-ins have to be installed on all clients and often require considerable amounts of time to load. An example for such a technology are Java applets. These are special Java programs that run within Web browsers. The browser is extended by a certain plug-in, the Java runtime environment (JRE). The JRE provides a security context called sandbox that shields the applet from interference with other applications and guarantees that only certain operations on the client machine are allowed, e.g., read and write only to a temp folder instead of the entire file system. Parts of the application can be loaded on demand from the server to keep initial start-up times and network traffic low. Besides these differences, Java applets are ordinary Java applications (section 5.2.3). Java classes can be written so that they act as applet or stand-alone application depending on one parameter. Advantages are minimal overhead for developing a Web client in addition to the stand-alone client and deployment across platforms, since JREs are available for nearly any operating system and browser. However, each applet runs in its own JVM (Java Virtual Machine) and each JVM requires a substantial amount of memory (>10 MB) which means several seconds to load even on fast client systems. Some companies also forbid usage of applets due to security considerations, although Java applets are commonly considered secure, especially if signed applets are used. In order to overcome some of these drawbacks and regain market shares lost to AJAX-, Flash- or Silverlight-based applications, Sun recently introduced JavaFX together with the runtime environment JRE 6 update 10. JavaFX allows developers to describe GUI elements with a compact declarative Java syntax and provides a scripting language to manipulate the GUI elements. It further brings powerful vector graphics capabilities together with media codecs for audio and video playback. The new JRE is leaner than its predecessors and allows to load only a basic set of classes and therefore faster starting times. Together they allow for rapid development of compelling rich Internet applications (RIA). Another way of enriching Web-based applications is called ActiveX, a technology developed by Microsoft. ActiveX controls run natively in Internet Explorer on Windows and Macintosh platforms, but controls that are developed for Windows which is the vast majority, do not run on Mac-

5.2. Client-side Access Technologies

intosh and vice versa. Capabilities of ActiveX controls are similar to Java applets and partially integrated with Windows, e.g., drag&drop from Windows applications to ActiveX controls. Differences are that ActiveX controls do not have a sandbox-like security mechanism. It can be defined that ActiveX controls have to be digitally signed and users get a warning before an ActiveX control is loaded. However, users can only choose whether to run the control or not and no security context can be specified. Macromedia’s Flash technology was originally designed as player for multimedia shows and animations available stand-alone or as plug-in to a Web browser. Adobe has extended these capabilities after the Macromedia take over, so that client application development has become much easier. Flash now offers a set of GUI elements, data binding to server-side data sources over XML and Flash Remoting to connect to an application server (J2EE, .NET or SOAP-based servers). Additionally, better audio and video codecs like h.264 were incorporated to enhance the multimedia capabilities especially for online movie clips. Flex is a server-side technology for application development that uses the Flash player as client. Flex applications use MXML, a Macromediaspecific XML dialect to describe GUIs, and ActionScript for client-side scripting and invoking of server-side methods. The MXML and ActionScript files are compiled into an SWF file (ShockWave Flash) and then transferred to the client that runs it using the Flash player. Further data is exchanged in the action message format, via Java RMI or Web services. advantages of Flash and Flex are a compact, fast-loading client-side plugin that is available for many platforms and advanced multimedia capabilities that enable the creation of sophisticated user interfaces requiring small bandwidth. Microsoft Silverlight was recently introduced to keep track with the enhancements of Flash, Java and HTML together with AJAX. Silverlight is available as a plug-in for major browsers and enables them to run rich Internet applications based on Windows Presentation Foundation. In version 2.0 due in second half 2008 it will be able to run the complete .NET 3.0 code. Version 1.0 is limited to displaying XAML GUIs that have to be programmed using script code. The open source project Moonlight is striving for bringing Silverlight functionality to non-Windows platforms based on Mono (see ”Mono” on page 381). A recent development that is not necessarily browser-based and shows that existing alternatives for client technologies are still not satisfying can be summarized under the term remote GUI. The idea is to use a platform, independent from the operating system, e.g., Java with its GUI APIs (like Swing or SWT), and implement a transparent network connection that transfers application events to the server and sends server responses back to the client. Advantages of this approach are low network traffic as the client interface is completely handled on the client-side, whereas the client

379

Flash and Flex

Silverlight

Remote GUI

380

5. Access Services

is relatively independent of the application version of the server, because the entire domain logic runs on the server. As most existing remote GUIs are Java-based, disadvantages are similar to applets, namely the substantial time required to load the JVM. This is a recent branch of GUI developments, so there is no standard yet, but quite a few startup companies offer first products. Examples are Altio Live, Asperon AppProjector, Canoo UltraLightClient and Droplets.

5.2.3 Rich Clients

Java, Java WebStart

AWT, Swing

In some situations, the best choice might be to build a stand-alone client that has no limitations regarding access of system resources. As there are abundant technologies available for implementing applications in general and GUIs in particular, we can only briefly discuss some examples for more or less platform-independent frameworks offered by different vendors. A wide-spread programming language for platform-independent development of client applications is Java, an interpreted language that needs a runtime environment to execute an application. Java offers a lot of rich features and is a proven platform, but also has some drawbacks as mentioned before. These are slow loading of the runtime environment (JRE) and consumption of substantial amounts of memory for small applications (>10 MB). A copy of the JRE is loaded for every single Java application which can soon lead to memory problems. In recent JREs (version 1.5, also known as 5.0) this issue is addressed by letting applications jointly use JRE class files in memory. With Java 6 update 10 there is an effort to modularize the JRE in order to keep the memory footprint small (< 5MB for typical applets). Another drawback is that central deployment of Java applications, i.e. installation of applications over network connections on demand, is not possible. Java WebStart enables administrators to make the latest client version available on the server. Users that have not yet installed this client have to download the classes. Once the classes are on the client, the application can be used as often as wanted without any further downloads necessary as long as the classes on the server are not updated. Thus, the application is started from within the Web browser, but then runs without the Web browser as stand-alone application. Sun is the main developer and founder of Java and supplies two class libraries for GUIs: the Abstract Windowing Toolkit (AWT) and Swing. AWT is a limited set of light-weight GUI components and considered relatively fast with respect to response time. Swing on the other hand provides a rich set of useful and optically appealing GUI components, but is considered rather slow as it does not use native API functions of the underlying operating system (OS). Instead, it renders GUI components itself using basic drawing functions. Therefore, GUI elements do not look identical to those used by native OS-specific applications which some users

5.2. Client-side Access Technologies

perceive as a disadvantage. However, in late 2008 there will be an updated version to Java 6 which brings native Windows rendering support (DirectX 9) to speed up Swing. IBM offers the Standard Widget Toolkit (SWT), a set of GUI components that can be described as Java wrappers around the native OS GUI components. Implementations exist for Windows, Linux GTK, MacOS X and a number of Unix derivatives. JFace is an extension to SWT that provides additional high-level GUI components like Wizards. Since it is based on native OS functions, SWT is considered to be comparably fast and the look and feel of SWT applications is closer to native OS-specific applications. IBM additionally provides an open source platform that has been derived from Eclipse, a Java integrated development environment (IDE) with abundant features and a large number of users due to its open source license model. With the version 3.0, Eclipse has been split into an application independent platform called Rich Client Platform (RCP, also called workbench) and IDE-specific functionality on top of it. The workbench uses SWT together with JFace and can be used by other application developers to build their own applications reusing functionality of Eclipse like the plug-in system or views. Eclipse has received tremendous attention, especially since IBM released Lotus Workplace 2.0, a comprehensive office suite built on top of the Eclipse workbench, now replaced by Lotus Symphony, which is an Open Office implementation using SWT. Microsoft's .NET platform is commonly seen as a direct competitor to Java, though it was not designed as a cross-platform development environment. Like Java, it provides a runtime environment and can be started directly from the Web browser using ClickOnce deployment (see ”Application server” on page 138). In contrast to Java, it is independent of the programming language used. This attracts many programmers that are not familiar with Java, but command C++ or Visual Basic. The .NET GUI components use native Windows GUI components and are therefore responding quickly. Version 2.0 is currently wide-spread. Version 3.0 extended 2.0 by workflow, messaging and new GUI functionality (WPF). The current version 3.5 adds many new features like language-integrated query (LINQ) and will be the future standard. An open source project called Mono provides a runtime environment for .NET applications for Linux, Mac OS X and other Unix systems. In current version 1.9, Mono offers class libraries for a certain subset of the large .NET API (e.g., ASP.NET WebParts). Still, Mono is nearly feature complete for .NET 1.1 and will offer full support for .NET 2.0 with some features of .NET 3.0 and even 3.5 like LINQ for in memory objects and XML data in version 2.0 expected in September 2008. Therefore, Mono already provides a robust framework for cross-platform development of .NET applications, especially when using C# and the available bindings to

381

SWT, JFace

Eclipse rich client platform

.NET

Mono

382

5. Access Services

Unix (e.g., Cairo library) and open source applications (e.g., ADO.NET for PostgreSql).

5.2.4 Desktop Integration In this section, we will discuss ways to integrate services of the EKI into employees’ desktop environments. We first have to explore which types of applications are used by typical EKI users and how they relate to knowledge work in order to understand the requirements for the integration of enterprise applications with desktop applications. Personal information management. The electronic desktop is the main interface office workers use for communication with their computer or, more precisely, the computer’s operating system. It is a metaphor that was created to help people to carry knowledge of how to do things in the physical world over to the electronic computer world. A first commercial product with a desktop, windows, icons, menus, and a mouse as pointing device was developed by Apple in the early 80s. Though PCs have seen dramatic changes since then as the table illustrates, the metaphor for user interaction is still largely the same (Table 5-2, Maier/Sametinger 2004). Table 5-2.

Early PCs and PCs today compared early PCs (1985)

PCs today (2008)

metaphor

desktop

object

document (file)

transformation

application (program)

interaction

WIMP (windows, icons, menus, pointing device)

software

limited number, primarily office applications

huge number, including multimedia and Web applications

resources

limited, e.g., 16 MHz CPU, 128 kB RAM, floppy disk drive (170 kB)

very powerful, e.g., 3 GHz dual core CPU, 4 GB RAM, 1 TB hard disk, DVD/Blu-ray (several GBs)

input / output

keyboard, mouse, small, monochrome monitor

keyboard, mouse, large color monitor, color printer, microphone, photo and video camera

documents

a few dozen or hundreds

several thousands

connectivity

isolated

permanent, fast network connection (100Mbit/s)

PC - user ratio

one desktop PC per user

one user often has a desktop PC, a laptop, a PDA and a smartphone

5.2. Client-side Access Technologies

The number of documents and resources have substantially increased and applications do not run isolated any more. PCs are always online and connected to servers and other PCs. That leads to the fact that the desktop metaphor is not as useful any more due to several problems: isolation: information is bound to specialized applications so that users have to remember which application they need before they can start the search for the information, e.g., “Have I read the piece of information in an email (email client), was it in an attachment that is in the file system now (file manager) or was it a link to a Web site and I have bookmarked it (Web browser)?”. Information is rarely reused across applications. loss of context: meta-data that often already exists in electronic form at some point of time, e.g., when an email with attachments is received, is lost when files are stored in the file system, e.g., who sent the file, the topic specified in the email subject and additional information in the email body. versioning: often, a document exists in multiple versions in the file system that either represent evolving states of the document or are representations of the same information in different file formats, e.g., doc, pdf, htm, rtf. That leads to the fact that often the file manager, the email client, the Web browser or other specialized programs are more often used for access to information as the desktop. Thus, a new metaphor has to be found. Several proposals have been made. Examples are (see Maier/Sametinger 2003 for an in depth comparison): Lifestreams: a stream of documents is used with temporal sequences instead of folders, Time-Machine Computing: an archive of the information history permits “time travels” through documents on the desktop, Presto: uses a uniform document model with no hierarchical arrangement, TheBrain: uses a semantic net to order and access resources on personal computers, Intranets and the Internet, Creo SixDegrees: a search engine for the desktop that uses search instead of browsing, Data Mountain: a collection of Web resources in a 3D space instead of bookmarks, TopicShop: collected Web sites are assessed and organized. Content of high quality is identified, Infotop: an OLAP-like, multi-dimensional view is used to access files, emails and Web sites. Dimensions are time, topic, location, person, process and document type. One- or two-dimensional views can be used for any combination of dimensions.

383 Problems with the desktop metaphor

Desktop alternatives

384

5. Access Services

Integrated file manager and Web browser

Today, the desktop is often tightly coupled with a file manager and Web browser (e.g., Windows XP integrates the desktop, Windows Explorer and Internet Explorer, the KDE Konqueror integrates file manager and browser). Examples for other file managers are Nautilus and Norton Commander clones like TotalCommander. Most widely-used browsers are Apple Safari, MS Internet Explorer, Mozilla Firefox and Opera. Other office software used quite often in daily work are (1) text processors, e.g., Corel WordPerfect, MS Word and OpenOffice Writer, (2) spreadsheet software, e.g., Corel QuattroPro, MS Excel and OpenOffice Calc and (3) presentation software, e.g., Corel Presentations, MS Powerpoint and OpenOffice Impress. These three types of applications are usually bundled in office suites, e.g., Corel WordPerfect Office, MS Office or OpenOffice and also provide integration between the applications to some degree. Desktop DBMS, e.g., FileMaker Pro, Lotus Approach or MS Access, are part of extended office suites. Another type of application central to office work today are Groupware clients, e.g., Lotus Workplace Messaging, MS Outlook and Novell Evolution (section 4.3.2, 290ff). Groupware contents can be stored on users’ PCs or centrally on a Groupware server. Advantages of storing contents centrally are the possibility to access them from any PC in the organization or even through the Web from any PC with Internet connection using the Groupware Web interface. These often-used software programs are supplemented by occasionally used specialized software. The creation of text documents with sophisticated layout, e.g., marketing material, or hundreds of pages, e.g., text books, requires specialized software for publishing. Desktop Publishing (DTP) software, e.g., Adobe FrameMaker and PageMaker, Microsoft Publisher or Quark XPress, are better suited for these purposes. For Web publishing, HTML editors are mainly used when export from office applications is not sufficient and there is no content management system in the organization. There are WYSIWYG (what you see is what you get) editors, e.g., Adobe Dreamweaver or MS Expression Web, for inexperienced or layout-oriented users and code editors, e.g., Macromedia HomeSite and CoffeeCup, that target professional or programming-oriented users. For small text files, e.g., configuration or log-files, it is useful to have a compact text editor instead of having to use a large memory-consuming text processor, e.g., jEdit, UltraEdit or VIM. Spell checkers are useful in order to aid users in creating texts and can be combined with grammar checkers and thesauri. Another often needed service is language translation. Current applications help to get a rough notion of the meaning of text, suggest translations and dictionaries provide some additional help. There are also a number of programs called readers or viewer for quick access to text documents without requiring the entire application which is only needed to edit and create the documents, e.g., Adobe Reader, GhostView or MS Word Viewer.

Office software

Groupware client

Publishing software

Other text-oriented software

5.2. Client-side Access Technologies

Handling of multimedia information becomes more important to office workers. We briefly summarize the most important types of multimedia applications. For images, one has to distinguish between vector and bitmap images. In the context of office work, vector graphics can be used to visualize technical or modeling information, e.g., workflows, organigrams or network architectures, or to illustrate text, e.g., culprit. For the former usage, applications like ConceptDraw, MS Visio or SmartDraw are well suited whereas the latter requires full-scale illustration software like Adobe Illustrator, CorelDraw and Adobe FreeHand. Sometimes, vectororiented animations are useful, e.g., supported by Adobe Flash. Bitmap graphics like photos and icons can be edited with products like Adobe Photoshop, GIMP or Corel Paint Shop Pro. Applications for easy management of large collections of image files, e.g., Google Picasa, Adobe Photoshop Album or ACDSee and quick viewing of images, e.g., IrfanView, JPegger or XnView, are often useful as well. Video clips can be watched with players like the Real Player, Windows Media Player or Totem. Sound files can be played with, e.g., Apple iTunes, WinAmp or amaroK. Those programs often also offer management functionality for audio files. Sound recording and editing software includes Audacity, SoundForge and WaveLab. A different type of audio information is encoded in MIDI (Music Instrument Digital Interface) files. Those can be seen as coded information whereas wave file formats represent audio data as noncoded information (section 4.4.2, 312ff). Finally, there are some useful tools that most office workers use sometimes including compression tools that compress files lossless into archives, e.g., GZip, PowerArchiver, Stuffit, WinRAR or WinZip. CD and DVD burning programs are also more and more important, e.g., Ahead Nero, k3b and Roxio WinOnCD. The number of different file formats is legion, and we cannot introduce even a fraction of them. However, there is a recent trend in file formats that is noteworthy. On the one hand, XML is used for an increasing number of applications as the primary internal format. On the other hand, pure XML files are not very efficiently stored, but can be compressed very well with existing compressing algorithms. Therefore, a number of applications store their data as ZIP-compressed XML data. This is especially true for both Open Office 2.x which uses the ISO standard OpenDocument (ODF) as well as MS Office 2007 which uses Office Open XML (OOXML), which was recently declared as ISO standard as well, but with several changes from the format used in Office 2007. Since office documents are compound documents these compressed file formats use an other advantage of the ZIP container. They store embedded objects as separate files in the container (e.g., JPEG images) so that they can be integrated very efficiently and do not have to be stored inline the XML document as base64 encoded byte stream. This causes tremendous efficiency improvements for

385 Multimedia applications

Software tools

File formats, OpenDocument, Office Open XML

386

5. Access Services

bitmap files as well as for embedded OLE objects from other office applications (e.g., an Excel table stored within a Powerpoint presentation). The latest enhancement of ODF in its upcoming version 1.2 is the integration of RDF-based metadata that can be used to semantically describe contents of the document referenced by an xml:id attribute. Table 5-3 gives an overview of important file formats for the applications discussed above. Table 5-3.

Overview of file formats

subgroup

Amounts of data

format

text formats

TXT, RTF, DOCX, ODT, CSV, HTML, XML

page description formats

PS, EPS, PDF, FM, PM6, QXD

other office formats

XLSX, ODS, PPTX, ODP, MDB, SDB

bitmap file formats

TIFF, BMP, GIF, JPEG, PNG

vector file formats

EPS, WMF, EMF, CDR, SVG

video file formats

AVI, WMV, RM, MOV, MPEG, DivX, XVid

audio file formats

WAV, MP3, RAM, WMA, AAC, OGG, AU, MIDI

compressed file formats

ZIP, RAR, TAR.GZ, ACE

The amount of data produced in organizations worldwide is astonishing. Estimations for the year 1999 are summarized in Table 5-4. Averages for newspapers, journals, magazines and newsletters are calculated per issue, assuming 300 issues per year for newspapers, 6 issues per year for journals and magazines and 52 issues per year for newsletters. However, there is even more informal information exchanged. A study for 2002 shows that the production rate is still increasing and estimates, e.g., 274 TB of data generated by instant messaging and 400,000 TB of data exchanged as emails (Lyman et al. 2003). Table 5-4.

medium

Produced data volumes per document type in 1999 (based on Asprey/Middleton 2003, 13) titles in thousands

average size

total size

968.7

8.3 MB

8.0 TB

newspapers

22.6

3.7 MB

25.0 TB

journals

40.0

8.3 MB

2.0 TB

magazines

80.0

20.8 MB

10.0 TB

newsletters

40.0

0.1 MB

0.2 TB

7,500,000.0

0.03 MB

195.0 TB

books

office documents

5.2. Client-side Access Technologies Table 5-4.

medium

Produced data volumes per document type in 1999 (based on Asprey/Middleton 2003, 13) titles in thousands

average size

total size

music CDs

90.0

66.7 MB

6.0 TB

DVD videos

5.0

4400.0 MB

22.0 TB

total

387

7,501,246.3

268.2 TB

Modern office systems are no longer stand-alone applications. Instead, they are integrated systems that often handle data from central data stores on Intranet or Internet servers in addition to local data. Therefore, intelligent synchronization or replication mechanisms are required that control data redundancy on servers, PCs and mobile devices (section 5.3.3). Synchronization can also be used to share documents in a team or workgroup, e.g., with the help of Groove Virtual Office or Novell iFolder. Plug-in technologies. A proven way for integrating additional services into desktop applications is using a plug-in system. The implementation requires interfaces defined by desktop applications, e.g., Web brokers, and software modules that implement those interfaces. Thus, the application can use these modules. For example in the Adobe Photoshop plug-in system, several hundred plug-ins are available for Photoshop that add new visual filters and effects. Some competing photo editing applications have also incorporated the system, so that the same plug-ins can be used. File managers are often used to access files for viewing or editing. Applications can be launched implicitly by opening a file with a file manager or the desktop, also called document-centric paradigm, instead of explicitly choosing an application and opening the document from there, also called application-centric paradigm. Users mainly need discovery and publication services in connection with the file manager, i.e. integrated access to files on their local computer, remote file servers and especially remote DMS and CMS including navigation and search. The Windows Explorer provides a large set of integration points. There are three possible integration points for plug-ins. New commands can appear (1) in context menus of folders or certain file types, (2) as new entries in the menu bar or (3) as icon in the toolbar. (4) Plug-ins can display an icon for the module as part of the directory tree as network drive. Examples for context menu integration are compression with almost all Windows compression programs (WinZip, WinRAR, AlZip), Adobe Acrobat document conversion and UltraEdit editing features. A static way to achieve similar results that also works for Linux KDE is the assignment of functions to file types via the options dialog. The menus and toolbar icons are handled equally in Windows Explorer. Modules that use this integration point are called file manager extensions

Synchronization

File manager plug-ins

Example: Windows Explorer integration

Context menu

Toolbar and menu

388

Network drive

WebDAV

Web browser integration, ActiveX

Netscape plugin API

Nested elements

5. Access Services

in Microsoft terminology. Examples for file manager extensions are Groove folder synchronization and Opentext Livelink Explorer. Integration as network drive is the most complex way of integrating to Windows explorer since users expect to get the same functionality as with local folders, e.g., context menus, drag and drop operation, searching. It is often used for DMS, e.g., Opentext Livelink or Windream, to access the contents of the DMS just like files on a shared network drive. Some manufacturers also use this mechanism to show the controls for hardware devices, e.g., Logitech Quickcam. A much simpler alternative to network drive plug-ins is the implementation of WebDAV on the server application (section 2.2.3, 105ff). Windows Explorer can connect to applications supporting the WebDAV standard by mounting a network drive. Many other file managers have the same capability. Thus, full read and write access is possible, but no application-specific context menus as it is with plug-ins. Web browsers need to display various file formats, especially multimedia and XML files that are required to facilitate GUIs for Web applications. Manufacturers cannot provide every desirable functionality on their own in a timely manner. Therefore, Web browsers usually offer plug-in systems that enable third-party manufacturers to build modules that fill the gaps. Examples for Internet Explorer (IE) plug-ins are Adobe Flash, Java JRE, and viewer for scalable vector graphics, XForms and other XML standards. Another essential plug-in for Web browsers is the Adobe Reader plug-in to enable PDF viewing directly inside the browser. Microsoft calls Internet Explorer plug-ins ActiveX controls (section 5.2.2). A problem for Internet Explorer plug-ins is that they have to use CLSIDs (class identifier) instead of mime-types to determine the plug-in to use. CLSIDs are the Windows standard to uniquely identify an application or library (DLL). That means that creators of Web sites have to prescribe which plug-ins are used instead of letting the user choose his favorite application to handle the given file type. This hinders open standards implemented by various vendors. Despite that, the big market share of the Internet Explorer and lack of a single plug-in system for other browsers which are cross-platform and therefore cannot use ActiveX led to the fact that many manufacturers of plug-ins provided only IE versions of their plug-ins or Netscape versions have not worked as well. Therefore, other Web browser manufacturers have agreed to use a single plug-in system, the Netscape plug-in API (NPAPI). In addition, they created an ActiveX control that is able to use NPAPI plug-ins within the Internet Explorer. That is a strong argument for plug-in vendors, so that browser plug-ins that work in any browser can be expected. One challenge for browser plug-ins for XML dialects remains. XML allows users to mix elements of various XML dialects within one document. It is for example possible to create an XHTML page that includes

5.2. Client-side Access Technologies

SVG-elements to graphically enhance headlines and XForms elements to let users enter messages to be posted to a newsgroup. XHTML is directly interpreted by the browser and different plug-ins display the SVG and XForms contents. However, if SVG and XForms elements are nested, e.g., when using SVG for form labels or when a form should be embedded in a SVG image, the page is not displayed correctly because one plug-in does not load another one. Therefore, modern browsers strive for implementing native support for these XML languages, e.g., SVG in Firefox 3, Safari or Opera 9. In the context of office products (text processor, spreadsheet and presentation software), users expect seamless access to documents stored locally, on file servers and in DMS. Additionally, users need integrated support for document publishing including versioning, format conversion and storage. MS Office is the most wide-spread office package. Since version 10 (Office XP), it offers a technology called SmartTag that was further extended in version 11 (Office 2003). SmartTags build on Word AutoCorrect functionality and offer a plug-in interface that has to be implemented to create own SmartTags. Every word typed in the text processor is parsed by the AutoCorrect function and gets verified for correct spelling. The SmartTag interface uses the event that a new word was verified to hook one or more modules conforming to the interface. It passes the word to modules which can decide whether it is a keyword or not. If it is found to be a keyword, it appears in the GUI as underlined with a dotted line. If the user moves the mouse pointer over this word, a menu pops up that offers functions in conjunction with the word. Functions associated with persons are for example send mail, open contact and schedule a meeting. There are also third-party products that integrate into MS Office using SmartTags. Ontoprise’s OntoOffice links words in the text document to concepts in an ontology. Thus, associated documents can be searched or associated concepts from the ontology can be displayed. The second Office extension point is the Task Pane, a side bar that can be used to offer additional functionality, like translations from internet services. Extensions that use SmartTags and Task Pane to integrate into line of business applications are called Office Business Applications since Office 2007. Office supports another plug-in system called Add-Ins. They consist of a document template that resides in the Office/Addins folder and is responsible for loading the module and creating the menus or toolbars and secondly either VBA code (Visual Basic for Applications) or a COM DLL (section 3.4.1, 219ff) that contains functions that are called by the menu entries or toolbar icons. Examples for applications that use this functionality are Adobe Acrobat to convert office documents into PDF, Groove Virtual Office to publish documents in Groove Workspaces or Opentext Livelink to enable users to load documents from and save to Livelink. A printer driver for an operating system has to conform to a pre-defined interface and is therefore similar to plug-ins. Users can print documents

389

Office software integration

SmartTags and TaskPane

Add-Ins

Printer driver

390

Groupware client integration

Applicationspanning plug-ins

5. Access Services

using a special printer driver to send them to an application, e.g., a DMS, instead of a physical printer. The printer driver gets a copy of the document that it can send to the application over the network, which stores it at a given location. Only writing to an application is possible, reading from the application is not supported. An example for a DMS offering a printer interface is Lacerte Document Management System. Conversion of documents to PDF is possible using Adobe PDF printer. Groupware clients are often used as central information access point for accessing emails, calendar items and contacts. One group of plug-ins for Groupware clients deals with supporting additional content types and information sources such as news in RSS (RDF Site Summary) format (e.g., NewsGator), additional messaging formats such as SMS (e.g., TeleMessage) or communication protocols for additional Groupware servers (e.g., SKYRiX ZideLook for access to Open Groupware from Outlook). A second group of plug-ins deals with a more comfortable or secure handling of the basic Groupware functionality by adding security functionality like encryption or digital signing for emails (e.g., PGP or FlexiSMIME) or automation of tasks like automatic compression of attached files (e.g., WinZip email Attachment Add-On for Outlook). Other plug-ins integrate Groupware clients with other enterprise systems like DMS (e.g., Livelink email integration client and Groove Virtual Office). A recent development is adoption of a plug-in system across application domains as it currently happens with the plug-in system of the open source platform Eclipse which is reused in many other open source applications.

5.3 Mobile Access This section deals with an important facet of knowledge work which is that knowledge workers need to increasingly work from home, client offices or while travelling. The section is structured into mobile computing in general (5.3.1), devices for mobile access (5.3.2) as well as mobile applications (5.3.3).

5.3.1 Mobile Computing

Characteristics

The term mobile computing denotes all electronically supported activities of non-stationary users. We will concentrate on employees that use electronic devices to participate in business activities or processes. There are some characteristics of mobile computing that differentiate it from stationary use of PCs: (1) access can happen any time and any place whereas EKIs traditionally were only accessed during business hours and from workplaces. An example is email access from many mobile devices.

5.3. Mobile Access

(2) Users can be located and information needs may vary strongly with the location. (3) Presentation of contents on mobile devices is different due to limited capabilities, e.g., screen size, resolution, and due to connection speed, as in most cases quick text information is more efficient than graphically rich information that needs several minutes to be downloaded. (4) Context information about environments other than traditional workplaces have to be considered. Utility of mobile computing for organizations can be broken down into the following factors. Employees can take part in the business processes and get information even if they are not at their workplaces or in company buildings. Without mobile access, workflows, e.g., waiting on approval by a mobile decision maker, often take weeks to be completed if the responsible person is out of office. Decision quality can be improved by providing required data at the right time and place. Data can be provided systematically depending on the current location of the user and vice versa, location data of the user can be used to improve business processes, e.g., ad-hoc optimization of routes for field sales force. However, mobile computing can induce high costs. Mobile devices have to be bought and maintained, mobile network services are more expensive than stationary ones and transfer times are long due to small bandwidths. Employees who benefit most from the new possibilities of mobile business are those that have jobs requiring activities at various places within and outside company buildings, those that travel a lot, often need current information to make decisions and those that occupy important positions within business processes and therefore have to be permanently reachable. Examples are field sales forces, consultants at customer sites, repair and maintenance teams, emergency doctors, as well as employees of procurement departments, coordination-intensive control stations, logistic units and other information-intensive positions. The use of location data is one of the most important characteristics of mobile business. Services are called location-based services if they use location data to enhance service quality. A number of different techniques can be used to locate the user. The only way to locate users with stationary devices is based on IP addresses and only reveals country, city and organization of users. The latter only works for devices with permanent Internet connections and locating is not exact (section 3.2.4, 191ff). In mobile networks, the cellular structure of the network is used to locate the user more exactly. The cell-of-origin method uses the cell a user is logged in to determine location. Depending on the region (urban or rural), accuracy of locating the user is between 0.1 km and 35 km. The time-of-arrival method is far more precise. It uses at least three base stations that are able to receive signals from the end-user device. The device sends a signal with a timestamp and each station measures the time the signal needs to get to the station. Thus, distances between user and stations are calculated with up to 50 m precision using triangulation. In contrast to

391

Costs and benefits

Target groups

Location methods, IP-based

Net-based locating

392

5. Access Services

Terminal-based locating, GPS

Location-based services

Higher level services

Table 5-5.

these two net-based methods, a group of different methods requires more “intelligence” for the mobile devices and is called terminal-based. The best known terminal-based method is GPS (Global Positioning System) which is controlled by the US military and uses a network of 24 satellites (plus four spare) to locate users. Again, at least three satellites must be available for triangulation. Despite the enormous distance (20,000 km), the method is very accurate. GPS allows locating up to a precision of 10 meters for public use and up to 3 meters for military use (section 3.2.4, 191ff). The European Union currently tests its own satellite navigation system called Galileo with 30 satellites that ought to be publicly available in 2006, but still are not operational in 2008. Disadvantage of satellitebased locating is that mobile devices need special, often expensive extensions to receive the signals. Additionally, it requires a lot of energy to send a signal that reaches the satellites which results in reduced operating time. Five basic types of location-based services can be distinguished (Reichwaldt 2002, 402f): Position acquisition solely denotes locating the user (“Where am I?”). The result is a position in geographical coordinates (longitude and latitude). Reverse geocoding is then applied to transform coordinates into an address (city, street and number). Geocoding transforms addresses back into geographical coordinates and is the first step if users initiate any address-related searches. Spatial search can locate other objects or users and calculate distances (air-line) based on geographic coordinates whereas routing also considers physical attributes of the region like streets, rivers and mountains to calculate distances and recommend routes to get to the target. Mapping generates a map of the user's local environment. On top of these basic services a number of higher-level services can be built. Examples are navigation systems, baggage locating, location-dependent notification services, e.g., advertising, traffic information, yellow pages and automobile roadway repair service. Table 5-5 gives an overview of these services and shows how precise locating must be in order to achieve broad user acceptance. Location based services and required locating accuracy accuracy broad acceptance

utility

locating occurrence

breakdown/emergency help 500 m

125 m

get help fast

user initiated

yellow pages

250 m

find points of interest nearby

user initiated

service

accuracy entry level

cell

5.3. Mobile Access Table 5-5.

393

Location based services and required locating accuracy

service

accuracy entry level

accuracy broad acceptance

utility

locating occurrence

traffic information

cell

cell

avoid traffic jams

user initiated

luggage locating

cell

cell

locate, route

user initiated

navigation system

125 m

30 m

navigate

every 5 sec.

notification

cell

125 m

get current local info

every 5 min.

billing

cell

250 m

differentiate prices

user initiated

A few case examples will further clarify how mobile computing contributes to organizational performance in various industries. A sales agent of an insurance company visits a customer who needs to insure his new car. The agent enters the car data directly into the insurance application on her notebook and looks up the rate online via an Internet connection established through her mobile phone. After the data is complete, she directly submits it to the company server that starts a workflow. Employees working in the nearest subsidiary of the insurance company immediately get the order, print the policy and send all necessary documents to the customer. After the repair process at the customer site is finished, the repair team enters the description of the case, measures taken and time needed. They submit the data via a GPRS connection to the coordination center and get new orders for the next assignment. Their current location, customers' locations, time and current traffic situation are taken into account. Every packet has a unique identifier (transponder, RFID) and all transport vehicles have mobile devices that make it possible to locate them. Thus, new services like packet routing, electronic letter of acknowledgement and just-in-time planning of parcel pick-up can be realized. A news reporter attends a press conference, takes photos and types an article directly on site. She inspects the pictures at her laptop and chooses the best fitting ones. She transfers pictures and text to the editorial office that compiles all articles into the online issue of the newspaper, so that timeliness of news is increased significantly. In case of an accident, an emergency physician is alarmed by the dispatch center via his smart phone and the address together with first information about number of persons and injuries are transferred as text messages to the smart phone, so that there can be no misunderstandings. The GPS-based navigation system helps him to quickly get to the site of the accident. After he has inspected the patients and has made a first diagnosis, he can send a request for more emergency teams, enter data into the smart phone describing the condition of the patient(s) and request or sug-

Insurance company

Service or repair team

Logistic service company

Newspaper

Emergency team

394

5. Access Services

gest a hospital. The hospital therefore has more time to prepare the intensive-care unit or call doctors that are on stand-by at home.

5.3.2 Mobile Devices The industry that develops and manufactures mobile devices is characterized by convergence of telephone and computer industry. Mobile telephones get more and more sophisticated and features are integrated that have originally been developed for PCs. Mobile devices are designed for different situations and usage scenarios which leads to varying product features. Characteristics. Main features can be categorized into communication features, technical device characteristics, programming capabilities and available applications (Table 5-6). Table 5-6.

Characteristics of mobile devices

category

feature

feature values

communication

voice telephony

supported / not supported

messaging

SMS / EMS / MMS / email

mobile Internet

WAP+WML / TCP/IP+HTML

memory

RAM: 1-4096 MB flash memory cards: supported / not supported internal flash disk: none / 1-64 GB hard disk: none / 20-500 GB

text input

12 digit keypad / QWERTY keyboard / touch screen with pen and handwrite recognition / voice

screen size, resolution and colors

from 1 inch,100x60 pixels, monochrome to 18 inches, 1920x1200 pixels, 24 bit colors

connectivity

PAN: USB, IEEE 1394, IrDA, Bluetooth LAN: Ethernet, WLAN, WiMax cellular: GSM, GPRS, EDGE, HSCSD, UMTS, HSDPA

operating time

from 60 minutes to 10 hours in use and from 3 to 15 days standby

programming capabilities

supported languages and libraries

Java MIDP / Java Personal Profile, .NET compact / Java Standard Edition, .NET, C++, …

available applications

PIM

address book, calendar, task list, notes

office

text processor, spreadsheet, presentation software

multimedia

image viewer / MP3 player / video player / image, sound and video recording and editing

technical characteristics

5.3. Mobile Access

Voice telephony was the first communication technique and provides synchronous communication, so that people could still share knowledge or make decisions while they were out of office. Additionally, a need for asynchronous messaging arose which was satisfied in a lean way with SMS (short message system) and richer with emails where message length is not limited. Information needs can be further satisfied with mobile Internet access as information from the Intranet or the Internet can be retrieved any time. It is important to note that access to the Intranet does not automatically mean access to all application services available there, as many of them have resource demands that only few device classes can fulfill. Technical device characteristics often determine whether using an application on a device is feasible or not. Enough available memory is usually a mandatory requirement whereas a limited keyboard with small keys makes the use of applications that require much text input only more difficult. Screen size and resolution determine how much and how information can be displayed. For some applications, this means that layout of the front end has to be adapted when porting it to mobile devices, other applications are not usable at all if they require a minimum resolution and size that is not provided. An e-book reader for example is usable on small screens if the text is formatted adequately, e.g., by using appropriate fonts, line breaks and image formats. If an e-book is, e.g., in DIN A4 format, users of small devices inconveniently have to scroll left to right and top-down to read the text. Connectivity options determine together with the availability of according base stations if and with which bandwidth users can work online. Without online access, they can use only the limited set of information on the device which has to be synchronized with a base station before and after offline work. Operating time is described in terms of standby and usage time. Short operating times limit usability of devices. Programming capabilities of a device are a significant factor for availability of applications. The more different devices a software developer can cover with one programming platform, the higher is the probability that this platform will be supported. Additionally, the more similar mobile platforms are to desktop platforms, the easier it is to port desktop applications to them. Therefore, Java MIDP (mobile information device profile, see also section 5.3.3) is currently the most supported platform as it runs on nearly all devices from small phones to larger PDAs (Personal Digital Assistants) and notebooks and its object model is similar to the Java desktop version (Java Standard Edition). Available applications also depend on device or operating system manufacturer, as the basic set of applications such as viewers, or PIM applications are usually packaged with the devices. Software providing office functionality, e.g., text processing, is mainly available for medium-sized mobile devices and bigger ones, e.g., PDAs and notebooks, though PDAs are often only suitable to briefly look-up information and correct or add a few words in existing documents. Multimedia applications are one of the

395 Communication

Technical characteristics

Programming

Applications

396

5. Access Services

key differentiators for mobile devices these days, although they currently are mainly oriented towards the consumer market and rarely needed for business activities. Mobile phone

Feature phone

Smartphone

Personal digital assistant

Device classes. Mobile phones are primarily designed for voice telephony and have SMS messaging capabilities. Besides that, some may also have WAP capabilities and a WML browser that work over a GSM or sometimes GPRS connection. These devices are typically small and have a monochrome screen with small resolution (e.g., 128x128 pixels) and small memory (e.g., 1 MB). Input possibilities are mainly a keyboard with 12 numeric and a few function keys. These keys are designed to type in telephone numbers. If any, these devices offer infrared connections. Feature phones are extended mobile phones with simple multimedia functionality and basic PIM support. They often support MMS and have a small color display (e.g., 176x220 pixels with 12 to 16 bit colors). Some of them also have an integrated digital camera with small resolution (VGA, or 1-2 Megapixel). WAP is a common feature and all products support at least one high-speed data service. Another feature common in this category is the possibility to download ring tones and background images that are stored in memory (typically 8 to 16 MB). Simple programs in MIDP format can be downloaded and used. Connection alternatives are at least USB and IrDA, some also support a couple of Bluetooth profiles. Smartphones are feature phones with full PIM and basic office functionality (text processor and spreadsheets), a larger display (e.g., 240x320 pixels with 16 bit colors and 2.8 inches) and larger memory (at least 32 MB) that usually can be extended via flash memory cards. Most have a digital camera with at least 2 Megapixel resolution. Advanced multimedia functionality like MP3 playback and video clip playback are also common. Smartphones regularly support several Bluetooth profiles. Internet connections are possible via TCP/IP so that standard HTML pages can be viewed. All products in this category support GPRS and EDGE or HSCSD and most recent products come with UMTS support, often even HSDPA. The integration of GPS functionality grew dramatically in 2008. Recent smartphones support at least the Java MIDP 2.0 specification and some of them support also advanced Java profiles like the PDA profile or even the Personal Profile or have .NET compact support. Smart phones are often bigger and heavier compared to feature phones (e.g., 120g, 109 x 56 x 16 mm). Input possibilities are often extended with a small keyboard that can be slid out or a touch screen with handwrite recognition. The main distinction between smartphones and Personal Digital Assistants (PDA) is that PDAs have larger screens (3.5 to 4 inches with 320x240 or 640x480 pixel) which leads to larger device size and weight (e.g., 150 g, 110 x 76 x 16 mm) and usually provide no support for voice telephony. Instead, they all provide powerful programming possibilities with Java Personal Profile or .NET compact and even more with a native

5.3. Mobile Access

compiler (mostly C++). Full PIM and nearly full office functionality is standard, as well as email clients and HTML browsers. PDAs have larger memory (64 to 256 MB) and support mass storage extension with flash memory cards. Standard input is via a touch screen and handwrite recognition or a virtual keyboard that is displayed on the touch screen. Some also provide a QWERTY keyboard. Connection to PCs is realized via USB. Many devices provide wireless connections via Bluetooth and IrDA more and more products additionally support WLAN. Some of them can be extended with or have built-in GPS modules, so that they can serve as navigation system. There are also PDAs that integrate voice telephony features and close the gap to smartphones. The border between Phone PDAs and Smartphones is only for categorization with pure PDAs increasingly vanishing. A feature of smartphones and PDAs that came up a few years ago is called Blackberry. It is an email push technology that sends emails directly to the device without requiring user action. Devices supporting Blackberry usually offer QWERTY keyboards to help users to create emails on the phone PDA or smart phone and sacrifice a large screen for it. Today, push email is no longer bound to devices from the manufacturer Blackberry, although they are still very popular. PDAs that get closer to notebooks are called clamshell PDAs, because they are foldable. Despite that, recent PDAs with full keyboard that fall into this category often have the keyboard sliding out of the back instead. They usually have a larger screen size (e.g., 5.0 inches) than other PDAs which leads to increased size and weight (e.g., 220g, 128x83x24 mm). To close the gap between notebooks and PDAs completely, Intel has claimed the invention of a new device category called Mobile Internet Device (MID) or Ultra Mobile PC (UMPC). They are tailored for multimedia and Internet usage and usually bring no voice telephony nor office software. They have screens with 4 to 7 inches and larger resolution (typically 800x480), a larger mass storage (20-80 GB) and use either touch screen or dockable QWERTY keyboards as input. Internet access is realized via WLAN mainly. UMPCs, as well as all devices in the following classes are x86-compatible and can therefore run the same applications as stationary PCs. Subnotebooks are lightweight mobile PCs with only a few limitations regarding display size, memory and computing power. They are substantially bigger than clamshell PDAs (e.g., 1400g, 270x188x36mm) and have 9-13 inch displays with WXGA (1024x600 pixels) or higher resolutions. They do not support voice telephony. The biggest limitations compared to larger notebooks are lack of periphery, e.g., internal optical drive, slower CPUs and keyboards that are not suitable for typing long texts. Subnotebooks with even more limited hardware resources (e.g., 8 GB flash disk instead of 40 GB hard disk) and low prices around 300€ are called netbooks. Connectivity to other systems is possible via IrDA, USB and Ether-

397

Phone PDA

Blackberry

Clamshell PDA

Mobile Internet Device, UMPC

Subnotebook, Netbook

398

Tablet PC

Convertible PC

Notebook

Desktop replacement notebook

5. Access Services

net. Some subnotebooks also provide an IEEE 1394 interface, Bluetooth or WLAN and have at least one PCMCIA (Personal Computer Memory Card International Association, a standard for expansion cards) slot, so that missing interfaces can be added as card solution. Tablet PCs are notebooks that can be controlled via pen-based input and handwrite recognition. They are similar to PDAs in the way that they have a touch screen, but in contrast to them are able to run standard PC software. They are nearly as small as subnotebooks (e.g., 1500 g, 324 x 220 x 22 mm) with display sizes ranging from 10 to 14 inches. CPUs, hard disks and connectivity capabilities are similar to those of subnotebooks. Convertible PCs (short: convertibles) are only slightly bigger than tablet PCs and provide both, a keyboard and a touch screen, so that one of these input devices can be used depending on situation and application. There are currently more convertible PCs than tablet PCs on the market and tablet PCs might be completely replaced by convertible PCs. Notebooks are bigger and heavier than subnotebooks (e.g., 2700 g, 340 x 275 x 35 mm) with displays usually ranging from 14.1 to 18 inch displays in the 16:9 aspect ratio wide screen format. XGA resolution (1024x768 pixels) is standard whereas larger ones have resolutions of 1400x1050 or even 1920x1200. With the additional space not only larger displays, but also faster CPUs, larger hard disks and larger keyboards are integrated, so that these notebooks are suitable for typing longer text passages and CPU-demanding applications. Notebooks usually have an integrated optical (DVD) drive with burning capabilities. Connectivity options include IrDA, Bluetooth, WLAN, USB, Ethernet and IEEE 1394. GSM or other cellular phone standards are not supported by default, but can be added as PCMCIA or ExpressCards. The same is true for GPS receivers. Desktop replacement notebooks are compact alternatives for desktop PCs and are not intended for mobile work in a narrow sense. These notebooks usually have more computing power and memory as well as large display (e.g., 17 inches) and powerful 3D graphic cards, but are much heavier and have short battery runtimes. Table 5-7 gives an overview of device classes and presents typical product examples for each class. Due to the short product life-cycles in the computer and mobile phone industry, many of them may soon be replaced with a successor, but nevertheless product pictures will be present in the Internet for a longer period, so one can get an impression of the devices. There are many other mobile devices available on the market and many of them integrate functions from other device classes. We are not discussing different types of MP3- and video-players, mobile gaming devices, photo and video cameras or mobile storage devices. These typically cannot connect wireless to companies’ Intranets to access emails or other sources. The market is always in motion, so that once established device categories like PDAs or tablet PCs vanish in favor of PDA phones and convertibles, whereas other types of devices like MIDs or UMPCs appear.

5.3. Mobile Access

Table 5-7.

399

Typical example products by device class (2008)

device class

example products

mobile phone

LG KP100, Nokia 1200, Motorola W205, Samsung SGHB100, Sony-Ericsson J210i

feature phone

LG Chocolate, Nokia 3500 classic, Motorola MOTOSLVR L9, Samsung SGH-M300, Sony-Ericsson W610i

smart phone

Apple iPhone, HTC Touch, LG KC550, Nokia N95, Samsung Qbowl, Sony Ericsson XPeria X1 and Toshiba G910

PDA

Asus MyPal A686, HP iPAQ hx2495, Palm T|X, HP iPAQ 211 (VGA)

PDA phone with keyboard and push email

BlackBerry 7100t, BlackBerry Curve 8300, HP iPAQ 910, Palm Treo 755p

PDA phone

E-Ten Glofiish X500+, HTC P6500, i-mate Ultimate 9502

clamshell PDA

E-Ten Glofiish M800, HTC Advantage X7501, Sharp Zaurus SL-C3200

mobile Internet device

Asus R2H, HTC Shift, Everun S60H, Nokia n810 and Samsung Q1

subnotebook and netbook

without optical drive: Acer One A110, Apple MacBook Air, Asus Eee PC 900 with optical drive: Lenovo ThinkPad X61, Sony Vaio TZ21, Toshiba Portégé R500

tablet PC

Fujitsu-Siemens STYLISTIC ST5112 and Motion LE1700

convertible notebook

Asus R1E, Fujitsu Siemens Lifebook T4220, HP Pavilion tx2500z, Toshiba Portégé R500

notebook

Dell Latitude D830, HP Pavilion dv6700t, Lenovo ThinkPad T61, Samsung P55-Pro

desktop replacement notebook

Acer TravelMate 7720, Dell XPS M1730, Sony Vaio AR71, Toshiba Satellite P300D-12F

5.3.3 Mobile Applications Based on the description of mobile devices, a number of characteristics limit development of mobile applications (Reichwaldt 2002, 44): small bandwidth, less processing power and small memory, limited output (screen size, resolution, no printer) and input capabilities (small keys, often no full keyboard), and

400

Mobile operating system

Symbian OS, Windows Mobile

Mobile frameworks, J2ME

MIDP

5. Access Services

high degree of variability in device capabilities and programming environments. The further up programming moves with respect to software layers from device-dependent machine code towards high-level programming interfaces, the higher is the number of users with different devices a programmer can reach with one programming platform. Therefore, high-level platforms are often favoured, although they may be slower and do not use all device capabilities. A first step towards standardization is the use of a common operating system. There are currently two important operating systems for mobile devices that are not x86 compatible, i.e. are not able to run Windows, Linux and other common PC operating systems. Windows Mobile currently is the dominant system within the PDA market whereas Symbian OS is leading the field of smart phone operating systems in Europe. The formerly dominating Palm OS has nearly disappeared and Research in Motio (RIM) with their Blackberry OS have gained some importance for Smartphones. Linux Mobile is predicted to play some important role in the future (e.g., as a result of the Google Android or the Linux Mobile (LiMo) platform) but has not yet gained significant market share. Both Symbian OS and Windows Mobile are similar in their capabilities and can be programmed using the programming languages C or C++ with OS-specific APIs. Main difference is the availability of additional applications. Developers usually use device simulators such as the Google Android Simulator to test and debug their applications before they are transferred to and tested on real devices. Unfortunately, simulators behave not 100% identical to real devices. A further abstraction is reached by using higher-level programming frameworks that are (ideally) independent of the OS. The most widespread framework on PDAs and smart phones is Java. The version used is called Java 2 Micro Edition (J2ME). The J2ME framework is broken down into a large number of specifications (Java Specification Requests, JSR) that are partly interdependent. The first distinction has to be made between the connected device configuration (CDC) and the connected, limited device configuration (CLDC). A configuration is a Java virtual machine (JVM) and a minimal set of class libraries and APIs providing a runtime environment for a selected group of devices. A configuration specifies a least common denominator subset of the Java language that fits the resource constraints imposed by the family of devices for which it was developed. The most frequently used configuration is CLDC which targets devices with 128-512 kB of available memory. On top of the configuration, a socalled profile supplies additional classes necessary to develop mobile applications. For CLDC, the mobile information device profile (MIDP) was created and is the most commonly supported profile for smart phones

5.3. Mobile Access

and even PDAs, although PDAs are not the targeted device group. For devices with at least 2 MB of available memory, the CDC would be suitable together with the foundation profile, but up to now there is hardly any implementation to be found. There is also a PDA profile with additional classes for the CLDC, but again there is no implementation up to now. In analogy to Java applets that run within a Web browser, the applications that run on top of MIDP are called midlets. MIDP provides features such as enhanced user interface, multimedia and game functionality, better connectivity, over-the-air (OTA) provisioning, and end-to-end security. Microsoft puts its .NET compact framework up against J2ME which provides good integration into the PocketPC PIM applications, but is only available for Windows Mobile platforms. Another possibility to create applications with platform-independent access is to use WAP micro browsers to access server-side applications. WAP uses WML2 as markup language that is based upon XHTML. The use of wireless profiled-TCP (WP-TCP) and WP-HTTP instead of WAPspecific network protocols leads to the fact that WAP2 browsers can directly connect to HTTP servers instead of using a WAP gateway. Nevertheless, using a gateway is reasonable, e.g., for transcoding content to fit small display sizes and memory of mobile devices. Otherwise, data transfer takes longer and users have to scroll both, horizontally and vertically to see the contents, or they have to use a special browser like Opera to scale down the contents which consumes considerable processing power. Several APIs such as the User Agent Profile or the Wireless Telephony Application (WTA) let WAP applications access user information stored on the SIM card or using device functionality such as initiating a phone call. With the availability of standard HTML Internet pages on most smartphones, WAP will diminish in the near future. Data exchange and synchronization. An important specialty of mobile applications is that they are usually limited versions of desktop applications or rely on a server to share data with other devices. Thus, they need data transfer and synchronization capabilities. A well-known standard for exchanging data on a high ISO/OSI level (e.g., calendar items or tasks) regardless of the concrete data structure is OBEX (object exchange protocol). It can be compared to FTP, but is optimized for mobile use and designed as push technology. It works on any reliable transport protocol like TCP, L2CAP (Bluetooth) or TinyTP (IrDA). It became popular through usage in Palm devices to exchange entries of PIM applications between devices (known as “beaming”). Principally, objects of any type can be exchanged so meta-data needs to be transferred in addition to binary object data. This can be done via a name header analogous to the MS-DOS file extensions (e.g., gif for an image, vcf for a contact or vcs for an entry in a calendar) or via an explicit type header that uses MIME types (e.g., image/gif, text/x-vCard, text/

401

.NET compact

WAP applications

OBEX, vCard, vCalendar

402

5. Access Services x-vCalendar). Thus, standards for the contents of object data are neces-

Synchronization

Proprietary synchronization software

SyncML

sary, in addition to OBEX. Examples for such formats are vCard (section 4.3.3, 292ff) and vCalendar (section 3.2.4, 191ff) that define structures of contact and calendar data respectively. For image and text data, common Internet standards like GIF, PNG, JPEG and ASCII text are used. Data of applications can be synchronized as well as applications themselves, e.g., to install them on the mobile device. Synchronization is the bidirectional exchange and transformation of data between two separate data stores in order to align their contents. Key tasks that have to be addressed by synchronization software are: data sub-setting and partitioning (only changed parts are transmitted), data compression (to keep network load low), data transformation (from one application-specific format into the other), transactional integrity (atomic, consistent, isolated, durable), detection and resolution of conflicting changes (e.g., if the same data record was modified on both sides), support for different transport protocols and multiple devices, authentication, authorization and encryption (for security reasons). There is a trade-off between number of synchronizations and currency of data that can hardly be addressed by synchronization software, but requires user decisions. Synchronization can be necessary between servers and end-user devices, e.g., Exchange server and Outlook client, or between end-user devices, e.g., PDA to PC file synchronization. Communication with servers seems to be currently moving away from proprietary network protocols towards Web services over HTTP. A similar development can be observed for device synchronization. It is still common for mobile devices that each has its own synchronization application installed, depending on the OS, e.g., Microsoft ActiveSync, Palm HotSync or Symbian Connect. Architecture and functionality of these applications are similar. Each provides a synchronization manager installed on the PC, handles the synchronization process and abstracts from the underlying network connection (USB, IrDA, Bluetooth, GSM, …). Additionally, an interface is specified and a set of implementations for PIM applications on the device are delivered. Third-party manufacturers can implement the interface to enable synchronization for their application. On the remote device, data can either be synchronized with the same application (maybe running on a different platform) or with an application of the same class, but from a different vendor. SyncML is standardized XML-based markup language for synchronizing data that seems to get more and more adopted in the industry, e.g., it is supported in Symbian OS 8. It distinguishes several types of synchronization like two-way synchronizations (based on flags or complete compare)

5.3. Mobile Access

and different one-way synchronizations where client data overrides server data or vice versa and either all or only new data is transferred. Conflict resolution is not addressed by SyncML, although a number of suggestions are made. Example suggestions for conflict resolution are duplication of the record, mix in of the changes if different fields are affected, overwrite data on one device or no synchronization of this record. Listing 5-2 gives an example of a SyncML file for synchronization of contact information between a Palm PDA and a PC. Upper level structure distinguishes a header part (SyncHdr) from a body part (SyncBody). The body part contains information about data source and target, data type, an identifier for the contact and contact data itself in vCard format. 1.1 ... ... ./contacts/ address.pdb text/x-vcard 2341 BEGIN:VCARD VERSION:2.1 FN:Herbert Schmidt N:Schmidt;Herbert TEL;WORK;VOICE:+49-345-4434124 EMAIL;INTERNET:[email protected] END:VCARD Listing 5-2.

Example of a SyncML document

The Add operation shown in the example defines that the contact data should be added to other contacts on the PC. Other basic operations supported are Copy, Replace and Delete. Additionally, there is support for transactions by using Atomic or Sequence elements to specify that operations must be processed all or none and in the given sequence.

403

404

5. Access Services

Case Example In order to make sales staff and pre-sales consultants as efficient as possible, they have been equipped with both a powerful notebook and a PDA phone with integrated GPS. Estefania was excited when she was equipped with the devices as her idea suggested in Global Industry Group’s employee proposal system finally got accepted. She mainly uses the PDA phone for managing contact data, keeping track of events and customer visits as well as navigation to the remote site. She can also check important messages directly on her PDA phone, no matter whether voice mail, SMS, MMS, instant messages or email. As she is an outgoing, communicative person, she always has her PDA phone with her. With push email and UMTS for high speed access or GPRS for rural areas without 3G network access, she is always connected to her inbox which works as main messaging center for all messages. Even faxes or voice messages from the wired telephone are routed there, thanks to unified messaging capabilities of the Groupware server. She is so glad that her boss, Susan, finally accepts that she can only supervise Estefania by her (impressive) achievements and not by the amount of time she spends sitting at her desktop in the office. It is the clients’ location where she needs to be or it does not matter any more where she is because she is always connected. Most information work that needs data input is done via the notebook that can be connected to the Internet either by an open WLAN available at the customer site or in public places or by connecting the notebook to the PDA phone using Bluetooth. Some applications like email and customer relationship management are accessible to the field staff directly through the Internet. An application-level firewall is used to authenticate incoming connections against the internal user directory, inspecting and filtering the secured (https) data stream and rerouting it transparently to the internal systems hosted in the demilitarized zone, often secured by a second firewall. Other systems are even more sensitive and can only be used by establishing a VPN tunnel to the corporate Intranet by using a one-time password generated by a smartcard which is considered even more secure. Such a VPN connection is also used in Estefania’s home office, with the difference, that the intelligent router that also connects the printer establishes the connection transparently in the background, so that for Estefania it feels like working in the office. She only has to put her notebook into the docking station and immediately is connected to the home office LAN and therefore also to the company Intranet so that the application portal used to access the important applications can be accessed. In order to keep Estefania’s children from connecting as well, 802.1x authentication is used between the home office router and the equipment plugged into it, so that only registered company devices can access the Intranet.

5.3. Mobile Access

Questions and Exercises 1. Why do users need portals as single-points of access if they already

have a Web browser as universal access application? 2. What is the relation between LDAP directory services, identity man-

agement solutions and portals? 3. What is the utility of personalization for the user of enterprise applications? 4. What are limitations of front-ends for enterprise applications that use only HTML and client-side scripting? Why does it make sense to use such thin clients instead of using rich clients? 5. What is the difference between applications implemented with Java Server Faces, Java applets and Java application started with WebStart? 6. What is the difference between desktop integration and function integration? 7. What is the utility of using client-side office applications to access server-side enterprise applications? 8. What impact does accuracy of location-based services have on mobile applications? 9. Why are there so many different classes of mobile devices? Which type of mobile device should be used in organizational settings to support employees doing knowledge work? 10.What impact do features of mobile devices have on mobile applications? 11.What are the main challenges for developers of mobile applications and how do application frameworks like .NET compact or J2ME contribute to deal with these challenges? 12.What is the advantage of mobile devices and applications that support synchronization using an open standard like SyncML compared to synchronization via vendor-specific exchange formats?

405

406

5. Access Services

Further Reading Collins, H. (2003): Enterprise Knowledge Portals. American Management Association, New York Collins was one of the first authors to bring together the technology of enterprise portals and ideas of the KM field. The book shows first examples of how to move into the direction of knowledge portals. Davydov, M. M. (2001): Corporate Portals and E-Business Integration. McGraw-Hill Education Although focusing a lot on the e-business side of enterprise infrastructures instead of knowledge-oriented applications, this book is one of the few resources that not only has the buzzword portal in its title, but really covers related technologies, integration and their strategic goals. Firestone, J. M. (2003): Enterprise Information Portals and Knowledge Management. Elsevier, Burlington, MA Like Collins, Firestone wrote an innovative book with respect to portal technologies and their applicability in the KM field. Mallick, M. (2003): Mobile and Wireless Design Essentials. Wiley Publishing, Indianapolis (USA) The book starts with an overview of mobile devices and network technologies and then discusses the development of mobile applications from several perspectives. It covers well-established technologies like WAP as well as recent developments like BREW, SALT and VoiceXML. Schiller, J. (2003): Mobile Communications. 2nd edition, Pearson Education Limited, (UK) Schiller paints a detailed picture of the technical basis of mobile communications. It is an excellent reference for the different mobile network protocols including their security features and partly even physical foundations. It can be seen as an encyclopedic treatise of mobile networks. Webster, S., McLeod, A. (2004): Developing Rich Clients with Macromedia Flex. Macromedia Press The authors have a great deal of experience with developing Internet applications with a variety of technologies reaching from DHTML over SVG to Flash. Although most of the book covers details about Flash and Flex programming, it also shows architectural issues and compares Flash with other technologies.

6 Conclusion

In chapters one to five, we presented the vision of an EKI and reviewed existing technologies that can be used to realize parts of this vision today. In this chapter, we will take up the introductory thoughts on knowledge work, knowledge management, KM instruments, IS architecture and EKIs and review results from both theoretic and empirical research projects. We start with the discussion of alternative methods for bundling EKI services to support knowledge work (section 6.1). An situation-oriented approach is suggested to overcome limitations that traditional workflowor process-oriented methods impose on employees. Then, we present a model of knowledge sharing suited to structure the basic steps as well as context factors that influence the knowledge sharing process (section 6.2). Afterwards, we take a look at the state-of-practice in KM by reviewing some of the findings of a comprehensive empirical study in large German companies (section 6.3). The dynamic environment and mobile way of doing knowledge work need flexible ICT support. Centralized systems are often not flexible enough to provide this support. We show in section 6.4 how peer-to-peer systems can complement centralized systems to provide better support for employees working in dynamic teams. Afterwards, we discuss the chances and risks of Web 2.0 from a knowledge perspective, which includes a discussion of the main characteristics of Web 2.0 (section 6.5). We end with a brief outlook on the future of EKIs from a technical and an architectural perspective (section 6.6). On completion of this chapter, you should be able to explain the concept of knowledge work situation, describe the steps of knowledge sharing and influencing factors, give an overview of the state-of-practice of EKIs, analyze advantages and disadvantages of client/server vs. peer-to-peer EKI solutions, describe characteristics of Web 2.0 and pros and cons for enterprises willing to adopt corresponding technologies for their EKIs and elaborate on future trends in EKIs.

Overview

Learning objectives

408

6. Conclusion

6.1 Situation-oriented Support of Knowledge Work Challenges for modeling knowledge work

Process-oriented KM

Knowledge work situation

Well-structured business processes can be modeled with the help of business process modeling methods and then can be supported by workflow management systems or generally process-oriented standard software as found in many application systems targeting enterprise resource planning. Knowledge work, however, consists in large parts of weakly structured processes that can only be marginally supported by traditional workflows. EKI are aimed at supporting knowledge work. We discussed a large number of concepts, tools and systems that can be combined to support knowledge work. Concepts are needed that guide the conceptual design of context-oriented, personalized ICT support for weakly structured processes connected to activities for knowledge sharing and learning as found in knowledge work. Under the headline process-oriented knowledge management, a substantial amount of research is currently undertaken to extend workflowand process-oriented concepts with knowledge-oriented concepts (see Hädrich 2008 for a detailed analysis and comparison). Examples are extensions to the architecture of integrated information systems (ARISKM), the business knowledge management (BKM) framework and its corresponding modeling method PROMET®I-NET, knowledge modeler description language (KMDL), and process-oriented methods and tools environment for KM (PROMOTE). Main extensions are the introduction of additional object types like knowledge object, i.e. topics of interest, documented knowledge, individual person, and skill as well as the introduction of model types like knowledge structure diagram, communication diagram and knowledge map. Ideas for concepts that have to be modeled in order to capture more detailed aspects of knowledge-intensive tasks have been implemented in tools for flexible workflow management, e.g., by anticipating alternative execution paths or leaving parts of the workflow unspecified until actual execution and by allowing modifications on workflow instances in order to incorporate unforeseen changes. Even though the added concepts describe a portion of the context of knowledge work, they are not suited to model the often unstructured and creative learning and knowledge practices in knowledge work and particularly their link to business processes. Knowledge work situations (KWS) are one approach to accomplish this based on a situation-oriented approach (Hädrich 2008), which is founded on activity theory. A KWS is a class of recurring situations in knowledge work defined by occasion, context, and mode resulting in knowledge actions. It describes a situation in which a knowledge worker can, should or must switch from a business-oriented function to a knowledge action. It does not represent any

6.1. Situation-oriented Support of Knowledge Work

409

situation of knowledge work but rather those that are most relevant for KM and shared by a number of individuals. Figure 6-1 shows the concept of KWS connecting on the one hand the business-oriented side, represented in three levels of generalization/specialization as value chain, event-driven process chain and sequence of tasks, and on the other hand the knowledge-oriented side represented by three levels of routinization as activities, actions and (automated) operations. process-oriented perspective

learning-oriented perspective

exploitation of knowledge level of motives

exploration of knowledge

value chains

activities refinement

routinization occasion context

level of goals

function

function

business processes refinement

level of conditions

task

task

mode

...

action action

action knowledge work situation

...

knowledge actions routinization

operation operation operation

workflows

Figure 6-1.

operations

Concept of knowledge work situation

In a process-oriented perspective, employees perform functions that are part of business processes. Simultaneously, they can be involved in an activity framing knowledge actions. Activities represent the learning systems that deal with the generation of new knowledge, which ideally but not necessarily needs to be related to the organizational core competencies. They can be focused on the business process or pursue a motive not related to the business process, e.g., an effort to build competencies. Thus, the knowledge action may have a more or less direct impact on completion of a business process. A business process can offer several occasions to learn and to generate knowledge related to the core competencies of an organization. Occasions trigger KWS and are associated with the tasks a business process is composed of. Occasions offer the opportunity or create the need to carry out knowledge-related actions. Thus, they represent the turning points where individuals switch from a mode of work that is primarily concerned with knowledge application to a way of working that is oriented towards the generation of new of knowledge which involves learning. It can be distin-

Occasion

410

Context

Mode

Knowledge action

Example

6. Conclusion

guished between primarily task-oriented occasions that trigger knowledge actions mostly contributing to the goals of the business process and having learning as a by-product and learning-oriented occasions that lead to knowledge actions mainly dedicated with learning and thus contributing stronger to the development of competencies. A KWS is not limited to generating knowledge, but may also include translating and applying knowledge created outside the KWS which in turn offers the possibility to generate knowledge. Context comprises all relevant information suitable to describe the actual situation of the worker. It includes both perspectives on knowledge work. Context information can be classified based on the dimensions person, i.e. all human-related information, process, i.e. information about (business) processes and tasks as well as about knowledge actions, product, i.e. information about the structure of knowledge incorporated by taxonomies or ontologies, productivity tool, i.e. all resources that are available for the knowledge worker as well as time and place. Since KWS are a concept to identify and structure technical support of knowledge workers, the context information suited to decide on and specify support with knowledge services is particularly relevant here. Mode classifies what actions can be performed. It can be described by four informing practices (Schulze 2004): ex-pressing is the practice of self-reflexive conversion of individual knowledge and subjective insights into information objects that are independent of the person, monitoring describes continuous, non-focused scanning of the environment and gathering of useful “just in case“-information, translating involves creation of information by ferrying it across different contexts until a coherent meaning emerges, and networking is the practice of building and maintaining relationships with people inside and outside the organization. Context, mode and occasion are means to specify a set of available, allowed or required knowledge actions. Examples for empirically identified actions are ad-hoc authoring, formal authoring, co-authoring, update, instruction of colleagues, submitting support calls, individual learning, person search, knowledge acquisition and invitation of people (Hädrich 2008, 335). Each of them can be detailed by a number of steps undertaken in order to accomplish a knowledge action. In contrast to the clearly defined sequences of tasks in the process-oriented perspective, there is no predetermined flow of actions. The learning about effective office product procurement in a consulting company can be described as an example of a KWS. It is related to the procurement process of the organization, more specifically to the function order product. The situation is triggered by the occasion that no suitable supplier is known for a certain product to be procured, e.g., a notebook

6.1. Situation-oriented Support of Knowledge Work

411

needs to be ordered that has unusual characteristics such as a pre-installed open source operating system and respective office software. The KWS is not limited only to this specific function because other business processes may also offer occasions that lead to actions contributing to the activity of learning about effective and sustainable procurement. Consequently, KWS may connect various business processes that are unrelated in the view of traditional process management. The facts that the case is urgent, that the procurement role typically does not need to have advanced technical skills related to hardware and software as well as the topics mobile computing, open source operating systems and office software are examples for context information. Privileges that govern access to technical systems are also part of the context. Relevant knowledge actions are the search for suitable suppliers based on written documents or the enquiry for colleagues that have skills related to the topics of the current situation. Knowledge services that might support the knowledge actions are email, phone, the Intranet search engine and publishing services offered by a DMS. Figure 6-2 summarizes the example described. activity: learning about effective and sustainable procurement

process-oriented perspective

exploitation of knowledge level of motives

learning-oriented perspective

exploration of knowledge

value chains

activities context: routinization case is urgent, role has low hardware and software skills, topics: “mobile computing“, “open source operating systems“ and “office software“, access privileges

occasion: refinement no suitable supplier is known occasion context level of goals

function

function

processes function &business business process: order product as part of the procurement process

level of conditions

task

workflows

Figure 6-2.

refinement

task

mode

...

...

action action

action knowledge work situation

knowledge actions: search for suitable suppliers, enquiry for skilled colleagues knowledge services: email, phone, Intranet search engine, publishing service

knowledge actions routinization

operation operation operation

operations

Example of knowledge work situation linked to procurement

From an EKI perspective, those KWS are of primary interest that can be supported by ICT. Depending on occasion, context and mode, it can be decided which knowledge services are suited to support the knowledge action in question. Therefore, KWS represent a promising extension of existing approaches for knowledge-oriented business process modelling. As not only one but rather multiple knowledge services are relevant in the

Modeling and ICT support

412

Context-dependent services

6. Conclusion

context of a KWS, e.g., in order to identify and interact with a domain expert, various different services such as skill directories and communication media need to be used, approaches are required that are suitable to flexibly compose integrated knowledge services based on more finegrained services. A service-oriented approach to EKI as promoted in this book is valuable as it offers possibilities for service orchestration and more flexible service composition. Context should be derived with as little user effort as possible. Open documents on the desktop, emails in the mailbox or the history of the Web browser can be used to determine context information. This can be enriched by data about the current function in the business process the user performs and data about actions that other users took in similar situations. Furthermore, awareness services may monitor current activities of other employees relevant in the KWS and thus be helpful in analyzing which cooperation partners are currently available or even engaged in similar business-oriented functions or knowledge actions respectively. Nevertheless, it is clear that it is still a big challenge to gather, relate and interpret context information automatically.

6.2 Context Factors for Knowledge Sharing Knowledge sharing is one of the central knowledge processes and despite abundant KM literature it is still not understood in detail how knowledge is shared and which factors affect success of sharing efforts. The challenge of analyzing knowledge sharing already starts with the term’s definition, as it depends on the underlying understanding of knowledge (see section 1.1) and the differing views of research disciplines. Synonyms for knowledge sharing are knowledge diffusion, dissemination, distribution, exchange and transfer. Knowledge flow and knowledge communication are other terms with a very similar meaning. The following definition regards knowledge sharing as a process embedded in a larger context: Definition of knowledge sharing

Knowledge sharing is the process of one person (source) (1) deciding to share knowledge, (2) remembering a portion of knowledge, (3) explicating it to contextualized information on a medium, (4) actively or passively transferring it to another person (recipient) which (5) perceives the information and (6) interprets it in the given context so that the knowledge is re-constructed and integrated in the person’s knowledge base. (7) Finally, newly acquired knowledge is evaluated by the recipient.

Discussion of the definition

This definition is narrow on the one hand since it does not include the occasion nor the application of the newly acquired knowledge as part of

6.2. Context Factors for Knowledge Sharing

the process but also broad on the other hand as passive transfer, e.g., by writing an article and publishing it is also treated as knowledge sharing. It explicitly focuses on knowledge sharing between individuals in an organizational setting and treats sharing between teams, organizational units or companies as a series of knowledge sharing activities on the individual level. In the following, the single steps of the knowledge sharing process are explained in more detail. (1) The process starts with the decision to share knowledge within the given situation to a known or anonymous recipient. The process can already fail in this early phase, e.g., due to lack of a good social relation to the recipient. (2) The source remembers a certain portion of knowledge that should be shared. This step is explicitly included in order to point out that remembering does not only lead to activating knowledge in the source and therefore slows down or prevents forgetting, but also is a reconstruction process that may change interpretation of knowledge by the source. Knowledge therefore is not a static entity that is stored and can be retrieved, but is reconstructed from memory dynamically and is interpreted before its explication in a given context. (3) Explication converts person-bound knowledge to contextualized information which can be perceived by other people. The smaller the portion of implicit knowledge, the easier is its explication. The implicit or tacit part of knowledge can be explicated only partly or not at all. Demonstration may be the only way to describe implicit knowledge whereas explicit knowledge can be explicated in oral or written form. (4) Once knowledge is explicated as contextualized information it can be transferred to a recipient using a medium. Oral transfer for example uses air as a medium, written uses paper. Both can be replaced by digital media. Direct transfer means clear addressing of a recipient, e.g., in faceto-face communication or by sending an email. Indirect transfer means that the recipient is unknown to the source or that an anonymous group of recipients is addressed, e.g., by writing and publishing a book. In this case, the transfer is intermediated which can lead to significant delays between explication and perception of the information. (5) Most important for perceiving information are auditive and visual senses. Haptic, olfactory and gustatory senses are not used widely in order to perceive complex information, although there are some exceptions like in chemical industry where smell and taste play an important role or for blind people who depend on their haptic sense to read texts. (6) Interpretation of perceived information leads to reconstruction of the explicated knowledge and largely depends on the cognitive state of the recipient. Therefore, the recipient obtains the meaning of the perceived information, though this also means that source and recipient will hardly have a 100% identical interpretation of the shared knowledge.

413

Decide

Remember

Explicate

Transfer

Perceive

Interpret

414

6. Conclusion

Evaluate

(7) Finally, the newly acquired knowledge is evaluated by the recipient. Previous and new knowledge have to be made consistent and potential conflicts have to be resolved. The evaluation result significantly determines the impact of this knowledge on the recipient’s future actions. Figure 6-3 summarizes the steps of the knowledge sharing process. Source, recipient, channel, message and context are the basic elements of knowledge sharing. These elements are represented by person symbols (source and recipient), gray arrow (channel) and envelope (message). knowledge context 1 decide

occasion situational context

3 explicate

2 remember

4 transfer

5 perceive

utilization 7 evaluate

transfer context

source context

6 interpret

utilization context

recipient context

organisational context relational context

Figure 6-3. Context of knowledge sharing Situational context

Knowledge sharing model

A number of context factors have to be considered in order to study knowledge sharing. When specializing on knowledge sharing within organizational settings, eight context areas can be distinguished, each of which aggregates a large number of factors already included in Figure 6-3. Particularly the occasion that leads to knowledge sharing as well as utilization of the shared knowledge by the recipient have to be considered. One can distinguish between pull and push occasions. In pull occasions, a person is actively requesting knowledge so that a potential knowledge sharing activity is initiated by the recipient. Push occasions describe situations where the source actively seeks people to share knowledge with, e.g., by telling a colleague how one has just solved a difficult problem. Typical occasions for knowledge sharing are decision support, where subject matter specialists share their knowledge with managers, professional training, where instructors teach course attendants and initial job training, where new employees learn rules and procedures from longtime employees. Four categories of occasions can be identified: social occasions like interaction between employees which is influenced by co-location, co-presence and communication frequency, organizationally-initiated occasions like regular meetings, post-project evaluations or job rotation, external learning occasions like interaction with customers and suppliers as well as attendance of conferences or trainings,

6.2. Context Factors for Knowledge Sharing

technically-mediated occasions due to awareness of colleagues’ activities and communication media for approaching others appropriately. Source and recipient context are two sides of the same coin since memorizing new knowledge and remembering existing knowledge are influenced by similar factors. Mental powers determine the source’s ability to identify and remember relevant knowledge and expressing it in a form so that the recipient can easily understand it. For the recipient, mental powers determine how well she is able to identify knowledge sources, memorize the newly acquired knowledge and how long she will retain it. Motivation is both needed for the source in order to hand on knowledge and for the recipient in order to be willing to learn which potentially can shake her view of things. It can be described by several factors. Type and characteristics of knowledge are a further group of factors influencing knowledge sharing, especially embedding it into other knowledge, share of explicit knowledge, affirmation and ambiguity. Transfer is influenced by how communication takes place. Personal transfer, e.g., writing an email, has to be distinguished from anonymous transfer, e.g., writing a blog entry. Synchronous communication such as face-to-face conversations differs from asynchronous knowledge transfer, e.g., creating a learning unit. Transfer can have different multiplicities: one-to-one, e.g., a phone call, one-to-n, e.g., a training session, or m-to-n such as a project debriefings. Media choice also causes differences. Within the relational context between source and recipient, the social relation is an important factor. Frequent contacts and a positive relation facilitate knowledge sharing. Trust in the source that she does not exploit the weakness of the recipient not knowing something as well as trust in the recipient not exploiting the knowledge to the source’s harm are other aspects of the social relation. Reliability, credibility and reputation are aspects of the social status. Expert status of the source facilitates knowledge sharing as well as organizational power. Finally, shared norms and common previous knowledge are beneficial. The informal culture of an organization as well as its formal structure can both hinder or facilitate knowledge sharing. An open atmosphere, task-orientation and failure tolerance are beneficial. Challenge and negative competition, a formal organizational structure and lack of feedback mechanisms may obstruct it. A long-term strategy might also be conducive to knowledge sharing. Finally, suitable application of management instruments like incentive systems can improve success of knowledge sharing. Application of knowledge is the final goal of knowledge sharing which becomes already obvious in our definition of knowledge with respect to influencing future action. Usage within a business process or project is one aspect. Usage to further share knowledge is another important aspect that is sometimes neglected, although several professions like teachers, instructors or consultants have specialized in these activities.

415

Source and recipient context

Knowledge context Transfer context

Relational context

Organizational context

Utilization context

416

6. Conclusion

6.3 State-of-Practice of Knowledge Management In the following, the state of practice of ICT-supported KM initiatives that has been investigated in an empirical study (Maier 2007) will be summarized in the form of theses that together describe activities concerning KMS in German organizations. Theses are based on the results obtained in a broad questionnaire conducted in 1999/2000, on qualitative findings from in-depth interviews with knowledge managers as well as experiences the authors collected in the last couple of years in consulting projects, conferences and discussions with researchers and practitioners. Theses are organized into four blocks: strategy, organization, KMS and economics. KM and KMS are increasingly implemented

Agreement about strategic relevance

KM initiatives as multidisciplinary effort

Strategy. Up to the last couple of years, more and more organizations showed interest in KM. So far, organizations are most successful in achieving rather basic KM goals in both, the codification and personalization side of KM, such as an improved access to existing knowledge or an improved communication and location of experts. More ambitious KM goals, such as turning implicit into explicit knowledge, or changing culture have been achieved to a much lower degree. Also, there has been a strong increase in the interest, the state of implementation and the usage of KMrelated ICT systems. Most organizations have installed an advanced Intranet infrastructure during the last years which they try to extend towards an EKI. Mostly large, knowledge-intensive organizations have invested in KM. Professional services firms and a number of pioneers in a variety of industries have been leading the way. As recent related empirical studies have shown, more and more small and particularly medium-sized organizations have started to evaluate the potentials of KM. Most organizations agree on the potentials of KM. The initiative quickly gains high visibility. Most KM initiatives report to the two highest levels of the organizational hierarchy. In many organizations, the executive board has pushed the approach. Organizations have high expectations towards KM which unfortunately can rarely be achieved within a short period of time. There is broad agreement over most empirical studies that KM is a relevant and important topic. The interviewees were also convinced about the positive effects of their KM initiative on business goals. But they admit to have difficulties in establishing clear, well-documented and measurable KM goals. The lack of a well-defined and (empirically) proven set of KM strategies is obvious as most organizations aim at a large number of different KM goals at the same time. This missing link and the measurement of the impact of KM or EKI on knowledge and business goals seem to be the most important challenges ahead. KM initiatives regularly comprise a strategically relevant combination of organizational and ICT instruments. Even though organizational instruments are the main drivers for a change in the handling of knowledge, it is

6.3. State-of-Practice of Knowledge Management

often ICT implementations that play the role of an enabler, a catalyst for the changes to take place as they visibly change work environments of participating employees. Consequently, multiple disciplines are required in order to implement KM successfully. In a substantial part of the organizations, KM is not embedded in a single functional area, but assigned to an interdisciplinary group. Also, many KM initiatives are split into at least two separate groups within an organization with frequently a large gap between their perspectives. These are enterprise communication, human resources and organizational design on the one hand and IT on the other hand. Regularly, marketing, R&D and strategy are also major players in the KM initiative. Most KM initiatives have their focus on knowledge flows between organizational units or groups of employees within the organization’s boundaries. Much less do KM initiatives aim at knowledge that crosses organizational borders. Neither do most organizations support the ICTsupported acquisition of external knowledge nor do they systematically profit from knowledge developed internally by selling knowledge products or services. Also, most KM initiatives only foster organization-internal work groups, teams, networks and communities whereas those collective structures that cross organizational borders are rarely systematically supported. Organizations have just begun to establish positions for key strategic alliance management that address these challenges at least for the most important liaisons to partner organizations. Organization. Today, large companies have a multitude of KM efforts working in parallel to tackle the problem. In many cases, several core groups start KM activities independently. Thus, in some cases even the various KM groups, teams and communities do not coordinate their efforts or even exchange knowledge which gives an indication of the complexity of the challenge. An EKI therefore should provide a platform for integrating individual ICT projects supporting KM that together could have much more positive impact on productivity of knowledge work. KM was implemented as a predominantly decentral approach leaving as much responsibility with decentral functions as possible. Responsibility for contents handled by EKI in most cases is shared between authors of knowledge elements and subject matter specialists. One of the most important goals of the implementation of EKI solutions is to increase participants’ ability to actively handle ICT supported knowledge, e.g., to publish knowledge elements and information about their skills, project assignments and the like and to react to activities of other participants, e.g., to answer questions and contribute to discussions. An EKI helps to decentralize corresponding KM tasks. A central unit, either a separate, permanent organizational unit or a project, frequently coordinates decentral activities.

417

Organizationinternal knowledge

KM as a set of independent activities

KM initiatives are decentral

418

6. Conclusion

Subject matter specialist as key role

Subject matter specialists are responsible for the majority of KM tasks surrounding EKI. They take on responsibility for one subject area or domain in the organizational knowledge base, help knowledge providers to document, link and organize their experiences, refine and organize their subject area and help knowledge seekers to locate expertise and knowledge elements. Subject matter specialists also play the role of “linking pins” for knowledge-related design tasks such as updating or mapping of ontologies as well as integration of knowledge sources into existing ontologies. There are also several organizations in which responsibility for KM tasks is not assigned at all. About a third of the organizations just assigned responsibility for basic tasks related to publication and distribution of knowledge, but did not pay equally high attention to what happens to it once it is documented and inserted into the organizations’ knowledge bases. In a number of organizations, important tasks, such as actualization and refinement of existing knowledge, quality assurance, deletion and archival of knowledge, are not systematically assigned. This might trigger a vicious circle in which participants use EKI less frequently because they do not find what they are looking for. Thus, investments in (contents of) EKI are cut which deteriorates quality of the contents. This reduces trust in the knowledge and in turn negatively affects usage of the systems starting the circle over again.

No responsibility for important KM tasks

Intranet and/or Groupware platform as foundation for EKI

KM functions are implemented, but not used

KM-related ICT systems lack integration

Systems. By now, almost all large organization have installed an Intranet and/or a Groupware solution which can be considered a basic ICT infrastructure and predecessor of an EKI. These platforms together with a multitude of extensions and add-on tools provide good, basic KM functionality. During the past couple of years, corporate Intranet solutions have been implemented to connect employees, to support the easy sharing of electronic documents and to support access to company information. Also, organizations have installed Groupware tools in order to support teams and to master the increasing complexity of organizational structure and processes along with advanced information and communication needs. Large organizations have already implemented many KM-specific functions as part of an EKI or in more specific solutions, such as content management, document management or customer relationship management systems. Many functions are not used intensively, in some cases due to technical problems, but mostly because they require substantial organizational changes. Therefore, there still seem to be considerable potentials when applying ICT to KM initiatives. In most organizations, a multitude of partial systems are developed without a common framework which could integrate them. Only recently, comprehensive EKI-like solutions gain market share. They offer extensive functionality integrated within one platform. Some organizations also build enterprise knowledge portals that at least integrate access to a num-

6.3. State-of-Practice of Knowledge Management

ber of organizational and organization-external ICT systems considered relevant for the KM initiative. Still, in most organizations functions and especially contents of KM-related ICT systems are largely dispersed in, e.g., messaging systems, document and content management systems as well as plain file systems. Each source of knowledge elements has its own basic taxonomy and home-grown meta-data concept, so that standardization of meta-data descriptions, extensions of taxonomies to ontologies and mapping of these seem to bear large potentials. Comprehensive EKIs are highly complex ICT systems because of (1) the technical complexity of the advanced knowledge services and of the large volumes of data, documents and messages as well as contextualization and personalization data that have to be handled, (2) the organizational complexity of a solution that affects business and knowledge processes as well as roles and responsibilities throughout the organization and (3) the human complexity due to the substantial change in the handling of knowledge that is required from the organization’s knowledge workers as EKI have to be integrated into their work environment. The majority of organizations relies on organization-specific developments and combinations of tools and systems rather than on standard solutions available on the market. The most important explanations for this finding might be two-fold. On the one hand, the market for EKI modules is confusing and dynamic. There is no leading vendor or group of vendors yet and interoperability with other EKI modules is still often difficult to realize because standardized Web services are not supported that much yet. On the other hand, organizations might fear that they loose strategic advantages if they exchange their home-grown organization-specific EKI for standard software that might not fit their needs as well. Generally, organizations handle a large variety of knowledge contents. Many organizations use modern KM contents, like employee yellow pages, skills directories, idea and proposal systems, lessons learned and good/best practices. Recently, organizations seem to have extended the scope of their EKI to include more types of internal knowledge previously unavailable to a larger group of employees. The biggest potentials seem to lie on the one hand in the systematic management of organizational knowledge processes that connect specific states of knowledge, e.g., personal experiences, lessons learned, best practices and, finally improved business and knowledge processes. On the other hand, external knowledge bridges the gap between the organization and its environment. Many organizations do not distinguish between these KM-related contents and more traditional contents of ICT systems, such as a broad view of all documents or the entire content of the corporate Intranet, data in data warehouses or transactional and communication data about customers and business partners. There is still considerable uncertainty in many organizations about what is or what should be considered a knowledge element.

419

EKIs are highly complex systems

Self-developed EKIs are prevalent

Diversity of EKI contents has increased

420

6. Conclusion

EKI is a major, long-term investment

Economics. Implementation of an EKI represents a major investment. A KM initiative and its support with ICT are long-term investments because they require a substantial shift in employees’ roles, organizational processes and often even a change of the organizational culture. Success is dependent on network effects. The more employees participate, the more useful the EKI will be and the more these people will profit from the solution. That is why individual ICT projects aiming at specific KM initiatives should be integrated on the basis of a comprehensive EKI. However, most companies apply KM-related ICT systems and concepts that promise quick returns-on-investment and are reluctant to commit themselves to a substantially higher investment to integrate their knowledge-related systems and especially to changes in work processes. Benefits of KM initiatives in general and an EKI implementation in particular so far are mostly determined by story-telling. In most organizations, this is the primary justification for the budgets allocated to the KM initiative along with references to similar activities performed by competition. The reason is that it is extremely difficult to measure knowledge directly. There are several promising approaches to the quantitative assessment of knowledge-related activities. They all require a fundamental shift in the organization’s management systems and in many cases organizations are as reluctant to massively change their management paradigms as they are in fundamentally changing their ICT infrastructures. The higher KM expenses per participant are, the higher respondents estimate the impact of a KM initiative on business goals. KM initiatives with a formal organizational design, but a decentral assignment of responsibility, a high rate of KM activity and systematic support of communities seem to be more successful than KM initiatives with a different organizational design. The relatively obvious tendencies in the case of the organizational design compare with a more uncertain picture in the case of ICT systems. An EKI seems to have no positive impact on business goals, but has to be combined with people-oriented and organizational instruments.

Success is assessed by story-telling

Organizational design is crucial for a successful EKI

6.4 Centralized versus Distributed Architectures Centralized EKI architectures

Many EKI solutions implemented in organizations and offered on the market are centralistic client/server solutions (section 6.3). In larger organizations, multiple servers can be accessed in the Intranet that offer specific services. EKI integration often happens with the help of an additional server that draws together data from various sources. The ideal architecture as presented in section 1.6.3, 75ff therefore applies the metaphor of a central knowledge server that integrates all knowledge shared in an organization and offers a variety of services to participants.

6.4. Centralized versus Distributed Architectures

An alternative to offer all EKI services on a centralized system is to implement them on distributed peers. There have been several attempts to design information sharing systems or even EKI to profit from the benefits of the peer-to-peer metaphor. This approach promises to resolve some of the shortcomings of centralized EKI, e.g., costs: to reduce the substantial costs of design, implementation and maintenance of a centralized knowledge server, external knowledge: to overcome the limitations of an EKI that focuses on organization-internal knowledge whereas many knowledge processes cross organizational boundaries and integration to desktop: to seamlessly integrate the shared knowledge workspace with an employee’s personal knowledge workspace. flexibility: semi-autonomous organizational units can easily create and share knowledge in temporary, dynamic networks of participants with the help of tools and ontologies that fit their domain, direct communication: knowledge is exchanged directly without central units that often act as unwanted filter to knowledge, acceptance: local storage together with efficient management of access privileges reduces barriers to provide knowledge that some central EKI solutions experience and is especially well-suited for expert-dependent organizations, e.g., doctors, lawyers, consultants. Implementation of a peer-to-peer-based EKI may either follow a pure peer-to-peer architecture or a hybrid architecture where peers are assisted by additional centralized servers or by super-peers (section 1.5.2, 56ff). The more coordination is required in a peer-to-peer system, as is the case in an EKI, the more likely it is that some kind of assistance by a server is needed to coordinate parts of the system. Principally, services to be implemented remain the same as in the centralized case, except those that target specific requirements of the peer-topeer architecture. Firstly, this is the need for discovery and communication with peers and eventually with centralized servers. Peers may dynamically join the network or go offline. Secondly, services to handle integration of the distributed knowledge base are needed, since it is ideally spread over all participating peers. Each peer has a knowledge repository that may be structured according to an individual ontology whereas central repositories managed by e.g., subject matter specialists might offer domain-specific ontologies and tools to map ontologies. Thus, the following services need to be provided in addition: Infrastructure services help locating peers, exchanging data with other peers and assuring security of the personal knowledge base. Servers may provide a connection to additional, shared data and knowledge sources and services for lookup and message handling that improve efficiency of the peer-to-peer infrastructure.

421 Benefits of peer-to-peer EKI

Additional services

422

The offline challenge

Offline data access with P2P

Product examples

Ad-hoc collaboration

P2P solution

6. Conclusion

Integration services that handle meta-data of the knowledge objects in the personal knowledge base and establish a personal ontology. Centralized replication services ensure synchronization of peers that sometimes work offline. The knowledge base comprises private, protected and public areas. Private workspaces contain information that is only accessible for its owner. Public workspaces hold knowledge objects that are accessible by all authenticated users. Protected workspaces contain knowledge objects that are accessible to a single or a group of peers that the owner explicitly grants access. A major problem of centralized architectures is supporting the offline scenario where mobile knowledge workers have no connection to the central server applications. Despite the propagated “always on” slogan, there is still a number of cases where working offline is the standard, e.g., at the customer’s site, in planes or trains. In that case, data might be not accessible and meta-data of downloaded data might be missing. Inferencing mechanisms or other advanced knowledge services are regularly not available offline as well. P2P systems could help to overcome these problems in hybrid scenarios where they act as offline clients for central server systems and are able to provide large parts of the centrally available functionality based on the limited data that was synchronized and is offline available. As synchronization is already one of the core concepts of P2P systems, it seems only natural to use it for synchronizing with central servers as well. Colligo Contributor is not a P2P system but a good example of providing an offline experience that is close to working online. It not only synchronizes documents stored in MS Sharepoint, but also makes meta-data and associated views on these documents offline available. It further provides strong desktop integration with MS Outlook and supports versioning. Microsoft Groove is an example for a P2P system that provides offline functionality for MS Sharepoint. As Colligo Contributor, it can synchronize contents back to Sharepoint and detects version conflicts if present. Another challenge for centralized infrastructures is a scenario in which users want to share a draft document to get feedback from their colleagues. In centralized architectures permissions are often defined by dedicated administrators in order to avoid chaotic permission structures. Further more it is usually avoided to assign permissions on single documents and to single users. Instead, they are assigned to groups of documents, e.g., folders or document libraries, and to groups of users. This makes ad-hoc collaboration impossible, because the reviewing colleagues need access and nobody else should be able to see the draft. This often leads to emailing the document around to collect feedback. With P2P software users can easily create own team rooms or collaboration workspaces and decide which documents to share in there as well as the participating team members that should have access. This is exactly the requirement for ad-hoc collaboration. P2P software might even offer

6.4. Centralized versus Distributed Architectures

423

online collaboration sessions where users can share their monitor with colleagues for jointly reviewing and enhancing documents. In addition to the technical aspect, there is also an emotional aspect associated in this scenario as users tend to experience a loose of control over a document when uploading it to a central server system which might even be located in a foreign country. In P2P scenarios the subjective control over the document is much higher. Comparison. An EKI may be implemented following predominantly a centralized or a decentralized approach. The decision about which approach may be applicable depends on the actual design and goals of an organization’s KM initiative. Table 6-1 compares both alternatives. Table 6-1.

Centralized and distributed EKI compared

characteristics

centralized EKI

distributed EKI

type of knowledge

stable, objectified, proven, generalized knowledge

ad-hoc, subjective, unsecured, specific knowledge

knowledge repository

centralized repository with an organization-wide ontology

decentralized repositories with ontologies reflecting shared contexts of groups

contents

(local) best practices, (approved) knowledge products, secured knowledge as well as ideas, experiences and individual contents

individual contents, ideas, results of group sessions and experiences, lessons learned, good practices

control

centralized

decentralized

goals

secure and reuse organizational knowledge

store knowledge according to individual requirements

KM orientation

technology-oriented KM

human-oriented KM

organizational design

hierarchy, departments

project teams, communities, ad-hoc groups

A centralized approach fits best with a technology-oriented KM strategy where securing and reuse of knowledge are primary goals. A centralized repository is built in which either all or only a specific part of the members of the organization contribute and retrieve knowledge which can be centrally controlled. In contrast, a decentralized approach seems to be most suited to support ad-hoc, individual knowledge and spontaneous interaction between individuals in groups. There is no centralized control and administration, neither of knowledge structures nor of technical maintenance and administration tasks. This approach best fits a human-oriented KM strategy where

Centralized EKI

Distributed EKI

424

Hybrid model

6. Conclusion

focus is on fostering social interaction and collaboration to jointly create innovative solutions. P2P software does not have to be an alternative for centralized server systems. They can also accompany them providing additional benefits. Documents can be jointly edited and developed in P2P workspaces and then uploaded to their final storage container on a central server where it is integrated into the organizational knowledge base and underlies the central lifecycle processes for document archival and deletion (see section 4.4.5). It is also easier for other users to get to know that a new document was created if it is stored centrally. A huge challenge is to electronically support this transfer from the P2P world into the central world so that existing meta-data can be reused and the user is prompted to add missing metadata. In order to accomplish that, users need to have an opportunity to map personal ontologies to centralized ones. It also requires a translation of meta-data that exists in form of complex data types into the existing data types on the central system.

6.5 Chances and Risks of Web 2.0 from a Knowledge Perspective A big success

Web usage

Ongoing debate about Web 2.0

The Internet in general and the World Wide Web in particular are big success stories in human communication which is reflected in many historical comparisons such as Gutenberg’s invention of the printing press. The World Wide Web was “invented” in 1991 by Tim Berners-Lee. The invention builds on older ideas, though, among others, hypertext and markup of text that have been brought together, simplified and thus made accessible to developers and users alike. Thus, the “Web”, as it has come to be called, is in its teens and has seen a stunning development since its inception. The Web has been used as information platform, communication medium, entertainment platform, business transaction infrastructure, for social networking and for building of knowledge, to say the least. Among other developments, the technological basis has seen an evolution with more recent technologies that have been absorbed and the patterns of usage of the Web have been continuously enriched. In the early 2000s, changes in Web usage were deemed so profound that they have sparked discussion whether it is time to point towards a new step in Web life, which Tim O’Reilly called the Web 2.0. There is much debate going on whether Web 2.0 really denotes a meaningful transition or whether it is merely a hype term that does not describe anything that has not been already there in the Web before. The two following polarized statements given by the two main protagonists of these two perspectives on the term Web 2.0 contrast this.

6.5. Chances and Risks of Web 2.0 from a Knowledge Perspective

“Web 2.0 is the business revolution in the computer industry caused by the move to the Internet as platform, and an attempt to understand the rules for success on that new platform. Chief among those rules is this: Build applications that harness network effects to get better the more people use them.” (Tim O’Reilly) “I think Web 2.0 is of course a piece of jargon, nobody even knows what it means. If Web 2.0 for you is blogs and wikis, then that is people to people. But that was what the Web was supposed to be all along.” (Tim Berners-Lee) Both statements contain a grain of truth as they view the term Web 2.0 from two different angles. Tim O’Reilly’s business is publishing and he noticed massive changes to his business with much more involved users, a change in sense-making and the business value of the Web. Figure 6-4 illustrates this using the indicator of the number of visits of individual pages of a Web site, called page impressions, as well as the number of page impressions that target user-generated content on the y-axis and orders the most popular Web sites in Germany in July 2008 on the x-axis. 6 out of 10 of the most popular Web sites have a more than 60% share of page impressions retrieving user-generated content and are therefore commonly considered as strongly buying into Web 2.0 facets, in this case, mostly social networking platforms with the most popular ones being specialized for school children (schuelervz) and students (studivz). 7

6 Page Impressions (PI) (billions) PI User-generated Content (billions)

5

4

3

2

1

Figure 6-4.

Popularity of Web (2.0) sites (data source: IVW 2008)

Wetter Online

meinestadt.de

Arcor.de

ProSieben Online

AOL

RTL.de

Spin.de

SPIEGEL ONLINE

MSN

MyVideo

MySpace

yahoo

Bild.de

meinVZ

Lokalisten

T-Online

mobile.de

StudiVZ

wer-kennt-wen.de

schülerVZ

0

425

Pro...

426

6. Conclusion

...and Con

Tim Berners-Lee views Web development from a more technology- and application-oriented perspective and, rightly, cannot see any “paradigmatic” or “revolutionary” change, but a continuous, evolutionary growth and refinement of development and usage of the Web. Figure 6-5 summarizes the most important characteristics that together differentiate a Web 2.0 site from a Web 1.0 site. In the following, we will review these main characteristics commonly attributed to the Web 2.0.

Characteristics of Web 2.0

data is “Intel inside” (control usergenerated content) network effects (harness collective intelligence)

end of software release cycle (develop continuously)

Webtop instead of Desktop (provide a rich user experience)

lightweight models (ensure cooperation & reusability)

software above the level of a single device (ensure deviceindependence)

Web as platform (comply with open standards) people (activating)

Figure 6-5. Web as platform

Network effects

technology (simplifying)

Concepts and characteristics of Web 2.0 (after O’Reilly 2005)

The Web, or network, as platform describes the notion of software being developed not to work in the environment of a single vendor’s operating system, but on the Web. This means that software is developed according to a system that is tied together by a set of open interaction standards that are not owned and issued by a single vendor, but are rather the result of agreements between major players in Web development and usage. This gives rise to Web 2.0 applications that are “executed” on several computer systems, e.g., they orchestrate Web services from different sites, i.e. machines, to present them to the user on yet another machine as a so-called mashup. Web 2.0 applications make the most of the intrinsic advantages of that platform. Harnessing collective intelligence means the services provided get more value with every increase of the number of people using it. Important phenomena that are typically related to goods and services that follow a pattern of network effects are the start-up problem, switching costs and lock-in. The start-up problem describes the effect that due to the low benefits of the service right after its start due to nobody using it, it is difficult to promote a new service. Switching costs mean the effect that once you have decided to use one service that profits from other people being also there,

6.5. Chances and Risks of Web 2.0 from a Knowledge Perspective

switching would not only require yourself to switch to a different service providing network benefits, but would require all other knots with which you are connected to also switch to a new service thus creating lock in effects. A typical example for a site benefitting from the “wisdom of crowds” is Wikipedia, where millions of users have created a wealth of articles on numerous aspects surpassing even serious traditional encyclopedias in comprehensiveness, let alone currency of its coverage. Another example are Blogs that profit from other people creating links towards Blog posts so that awareness about an article and in turn about its author is increased just by users themselves, without any agent or publisher being required to promote the user. This calls for providing user-generated content and services in a form that allows remixing by others and thus sparks an “architecture of participation”. User-generated content is the single most important asset in typical Web 2.0 applications which has been termed as “content is king”. Several start-up companies have quickly made a fortune by attracting large numbers of users to provide content and then selling off to one of the big industry giants such as Google or Microsoft. The most prominent example is Youtube, founded in February 2005 and launched in December 2005. The increase in usage followed an enormous growth pattern with 100 million clips watched a day, and 65,000 new clips a day in fall 2006. The company, consisting of still just 60 employees, was sold off to Google in November 2006 for 1.65 billion US$. Some Web 2.0 applications even consume and remix data from multiple sources. However, strictly speaking, a plethora of data without users coming back to the site is worth nothing. So, it is data plus (returning) people that drives a Web site. Interaction with the Web used to be strongly limited when compared to a Desktop application. In the beginning of the Web development, user experiences could be best described as a page metaphor, i.e. they were simply determined by scrolling through a Web page and clicking on links to open new Web pages. With the advent of AJAX, it has become possible to create Web sites that allow for rich user experiences, i.e. a look-and-feel similar to Desktop applications. AJAX stands for asynchronous Javascript and XML and denotes not a new technology, but a bundle of technologies such as XHTML, Cascading Style Sheets (CSS), the Document Object Model, asynchronous data transmission between Web server and Web browser by XMLHttpRequest and Javascript to arrange those technologies (see section 5.2.1, 374ff). Software in the Web 2.0 time is continuously developed and consequently in a permanent Beta status. Software is therefore delivered as a continually-updated service with no versioning or release anymore. This puts the open source motto of early and frequent releases to an extreme. The reason for this is that it is meant to attract users as testers and codevelopers of the software early on which gradually and quickly gets bet-

427

Data is “Intel inside”

Webtop instead of Desktop

End of the software release cycle

428

Software above the level of a single device

Lightweight models

Evaluation

6. Conclusion

ter the more a company is successful also to harness collective intelligence for the development of its software. Many people nowadays do not only have a PC at their disposal, but also other computers of varying power and sizes. These can be laptops, personal digital assistants, gaming devices, music or video players, Blackberries or smartphones. They all share the ability to connect to the Web, either directly or chauffeured by a PC. Thus, one aspect of Web 2.0 is also accessibility of contents and services from varying types of devices. The single most important factor is that many people tend to have mobile devices with them almost all of the time. Easy-to-use applications that bridge the gap between Web backend, PC and mobile devices thus are well suited to create interesting business models. The most prominent example for a successful integration beyond individual types of devices is Apple’s combination of iTunes and the iPod that allows for seamless integration between the music player and the backend music store responsible for much of Apple’s resurgent reputation in the last couple of years. Web 2.0 stands for the opposite of heavyweight programming models typically used for application design in intra-organizational settings. Web 2.0 applications are loosely coupled systems, that are designed for “hackability” and remixability and the single most important innovation is in assembly, not in creating new entire systems in an isolated way. In terms of content, Web 2.0 suggests to give up on control and tight coordination, but syndicate content to whoever is willing to reuse it. Reflecting on these facets, it does not really matter whether one views these phenomena as the ultimate goal that has been there when designing the Web in the first place (Berners-Lee) or a revolution on the Web (O’Reilly). The developments suggest that it does not make sense at all to think of systems handling knowledge as complex, centralized, integrated systems that are built in isolation from other systems. It is important to note that the architecture of an EKI is not intended to support this centralized model, but rather should be viewed as a collection of services on different levels that are required in order to foster the handling of knowledge in businesses and organizations. The Web 2.0 metaphor would help to argue that these services should be picked from different sources and assembled to support knowledge maturing processes. As in Web 2.0 the mashup of formerly isolated web sites generates a new application that sometimes becomes more beneficial than it’s underlying services we argue that integration of enterprise applications in EKI to combine their features and contents brings more benefit than introducing another isolated application. Also, there are strong arguments for the public release of all knowledge that is not mission-critical in order to profit from the benefits of having collective intelligence comment, change or improve it to the advantage of its issuer while at the same time improving the reputation of the organization that shares in the increasing community that subscribes to an open

6.6. The Future of Knowledge Infrastructures

429

content model as much as it used to subscribe to open source before. Table 6-2 summarizes selected chances and risks for organizations posed by the Web 2.0 viewed from the perspective of knowledge in a table of arguments. Table 6-2.

Chances and Risks of Web 2.0

Con-Risk

Pro-Chance

difficult data protection due to opening of company data

easy access to data (wishes, trends) from customers and partners

Web 2.0 applications introduce new and more complex security risks

company applications benefit from complementary contents & services on the Web

employees waste time by creating and reviewing content of eventually low quality

employees gain time by efficient internal communication

accelerated spread of bad reputation e.g., by customers

more accurate perception of own image and appropriate reactions

head hunters get access to company's experts

efficient recruiting of personnel

faster transport of information to competitors

faster transport of information to customers, partners and stakeholders

can cause legal disputes

public exchange of opinions creates trust and credibility

changing customer behavior may lead to loss of business

flexible adoption of new technologies may create new business

6.6 The Future of Knowledge Infrastructures Numerous technological trends influence the development of EKI. Many concepts, tools and standardization efforts concurrently take place on various layers of the architecture. This makes it impossible to reflect current developments for all services touched in this book. Thus, we will limit our considerations to the areas of network infrastructures, devices for accessing EKI, integration technologies and the market for comprehensive EKI solutions. Besides the ever increasing size of network bandwidth, e.g., GBit Ethernet to the workplace, and usage of fiber instead of copper cables, there is also a large number of emerging standards, especially for wireless networks. Even before Bluetooth, WLAN and UMTS have been widely adopted, the next generation of standards are under development. Exam-

Network infrastructures

430

Always online

Access devices

EKI accessed by variety of devices

Data sources

Integration

6. Conclusion

ples are nearfield communication (Zigbee) for short ranging and quickly established connections, ultra wideband (UWB) as cable-less replacement of USB and WiMax (Worldwide Interoperability for Microwave Access) for long-reaching high bandwidth connections. The trend clearly is towards mobile workers being always online, connected to a world-wide inter-operable network of networks that permanently offers personalized services in the context of time, location and the states of co-workers and connected servers. Seamless hand over between networks using different protocols is a substantial challenge here. Mobile devices undergo an ambivalent development. On the one hand, there is a strong convergence between device classes. PDAs implement UMTS communication means and mobile phones implement PIM functionality. Both device classes get enhanced with rich multimedia capabilities formerly found in specialized devices like MP3 players and digital photo cameras. On the other hand, there is an increasing number of devices that cannot be sorted into an existing category. Blackberry email push technologies result in new specialized devices. The number of different notebook classes is increasing in both directions to desktop replacement and mini-sub-notebooks. Both, convertible notebooks and PDAs with VGA screen resolution close the gap to ebook readers. Increasing sizes of disc, flash and main memory as well as faster CPUs enable new applications and allow mobile workers to have important data always with them. Enterprise information and communication infrastructures are still focussed on the desktop PC that is connected to an organization’s Intranet. The challenge will be to provide appropriate access to personalized services offered by whatever system run by the organization or by whatever provider in the Internet, no matter with what type of device via whatever network connection a mobile knowledge worker accesses the organization’s EKI. Generally, integration of data sources poses an increasing problem to the effective handling of documented knowledge in organizations. EKI solutions need to connect enterprise resource planning systems, document management systems, content management systems, e-learning platforms, data warehouses and business intelligence tools, product data management systems, CAD tools, project management tools, customer relationship management systems, visualization tools etc. The problem is that most, if not all of these tools and systems implicitly attempt to play the role of the leading provider of documented knowledge within a certain context, e.g., a project, a customer contact, a product, a process or a learning situation. Standardized languages and corresponding tools could broker semantics in the form of ontologies back and forth between these tools and systems. However, semantic integration is time-consuming and costly and thus limited to application areas where reuse of knowledge objects is highly likely to pay off. Therefore, we have some way to go until we will

6.6. The Future of Knowledge Infrastructures

see smart, interconnected services in an EKI that truly benefit from the fast developments in the hardware sector and the large amounts of data collected in heterogeneous applications. Standardization of languages that help to semantically connect data sources and applications, e.g., based on Semantic Web technologies, point in the right direction, though they suffer from their tremendous complexity. Furthermore, they need to be integrated with concurrent developments in the software sector. In the last few years there has been a clear trend towards highly modular infrastructures with decoupled systems that are integrated with dedicated middleware systems. The hype around Web services and service-oriented architectures accelerates this development. But when the application of Web services and Semantic Web technologies leaves prove-of-concept scenarios and moves towards real-world scenarios, it becomes apparent that complexity is not entirely mastered, but to some extent only relocated. Composition of Web services to form processes remains as much a substantial challenge as semantic description of Web services and the data exchanged between applications. Major vendors seem to understand the need for comprehensive, yet modular systems and have supplemented their offerings to provide the whole range of modules from application server to identity management and Web service composition. Recently, the market for centralized EKI solutions has seen a number of strategic alliances, mergers and acquisitions. Many vendors of KM tools have vanished from the market or their technologies have been integrated in product offerings of major software companies such as IBM or Microsoft, or of leading vendors of EKI technology, such as EMC or Open Text. However, there are still many small companies offering innovative tools that pose substantial challenges with respect to requirements such as scalability, integration with other application systems and platforms as well as security. Many innovative tools only target small groups of users or single users, e.g., in the case of some visualization tools. Most EKI solutions so far are comprehensive, server-based, organization-wide solutions that target large organizations with hundreds of knowledge workers. Peer-to-peer EKI promise to resolve some of the shortcomings of centralized KMS, however, major challenges still lie ahead until peer-to-peer systems can truly be called EKI and can be used to support the still growing share of users involved in knowledge work. Consequently, future EKI solutions might attempt to combine advantages of peer knowledge workspaces that directly exchange knowledge on the one hand with servers that help to achieve the advantages of integrated and quality assured partial organizational knowledge bases on the other hand. Many organizations have undergone substantial reorganization during the last ten years when they exchanged proprietary, unintegrated solutions for standard ERP systems. Both, horizontal and vertical integration of

431

Market for EKI solutions

Combination of centralized and peer-to-peer architectures

ERP as one pillar of ICT infrastructure

432

EKI as the other pillar

6. Conclusion

structured data stored in relational databases has substantially improved transparency of business transactions in organizations, increased data quality and flexibility of the organization’s business processes and reporting system, just to name two. Thus, ERP solutions are one important pillar of information and communication infrastructures in many business organizations. What is left is the integration of semi-structured data dispersed in numerous servers and individual PCs, largely unintegrated and consequently hindering knowledge work. This sort of data is typically addressed by KM solutions. The implementation of KM technology in organizations has entered a new stage. It is not anymore the quest for the best individual tool targeting a specific KM problem that organizations should engage in. Organizations should now systematically build a second pillar in their information and communication infrastructures. EKI focus integration of valuable knowledge elements needed in weakly structured knowledge processes as much as ERP solutions targeted integration of business data needed in well-structured business processes. Ultimate goal of implementing EKI in organizations is to substantially increase productivity of weakly-structured knowledge work as much as ERP and other process-oriented systems have increased productivity of well-structured data and service work.

Bibliography Abiteboul (1997): Querying Semi-Structured Data, Proceedings of the 6th International Conference on Database Theory, pages 1-18, Springer, London, UK 1997 Alavi, M., Leidner, D. E. (2001): Review: Knowledge Management and Knowledge Management Systems: Conceptual Foundations and Research Issues, in: Management Information Systems Quarterly - MISQ, Vol. 25, No. 1, 2001, 107-136 Alonso, G., Casati, F., Kuno, H., Machiraju, V. (2004): Web Services. Concepts, Architectures and Applications, Springer, Heidelberg Amberg, M., Holzner, J., Remus, U. (2003): Portal-Engineering - Anforderungen an die Entwicklung komplexer Unternehmensportale, in: Proceedings of the Wirtschaftsinformatik 2003, Medien, Märkte, Mobilität. Physica, Heidelberg, 795-817 Argyris, C., Schön, D. (1978): Organizational Learning: A Theory of Action Perspective, Reading (MA) 1978 Asprey, L., Middleton, M. (2003): Integrative Document and Content Management: Strategies for Exploiting Enterprise Knowledge, Idea Group, Hershey Austin, J. (1962): How to Do Things With Words. Oxford: Oxford University Press Bafoutsou, G., & Mentzas, G. (2002): Review and Functional Classification of Collaborative Systems. International Journal of Information Management, 22(4), 281-305. Barley, S. R., Kunda, G. (2001): Bringing Work Back In, in: Organization Science, Vol. 12, No. 1, 2001, 76-95 Barney, J. B. (1991): Firm Resources and Sustained Competitive Advantage, in: Journal of Management, Vol. 17, No. 1, 1991, 99-120 Baumgartner, P., Payr, S. (1994): Lernen mit Software. Series Digitales Lernen, Österreichischer Studienverlag, Innsbruck Berners-Lee, T., Hendler, J., Lassila, O. (2001): The Semantic Web. A new form of Web content that is meaningful to computers will unleash a revolution of new possibilities. In: Scientific American, May 17th, URL: http://www.sciam.com/article.cfm?id=thesemantic-web, last access: 2008-11-01 Chase, R. L. (1997): Knowledge Management Benchmarks, in: Journal of Knowledge Management, Vol. 1, No. 1, 1997, 83-92 Christ, O. (2003): Content Management in der Praxis. Erfolgreicher Aufbau und Betrieb unternehmensweiter Portale, Springer, Berlin Clark, R.C., Mayer, R.E. (2002): E-Learning and the Science of Instruction. Proven Guidelines for Consumers and Designers of Multimedia Learning, San Francisco, JosseyBass/Pfeiffer Clemens, P., Northorp, L. (1996). Software Architecture: An Executive Overview, CMU Software Engineering Institute Technical Report, Carnegie Mellon University Codd, E. F., Codd, S. B., Salley, C. T. (1993): Providing OLAP (On-Line Analytical Processing) to User-Analyst: An IT Mandate, E. F. Codd & Associates. Daconta, M. C., Obrst, L. J., Smith, K. T. (2003): The Semantic Web. A Guide to the Future of XML, Web services and Knowledge Management, Indianapolis (IN), USA

434

Bibliography

Daft, R. L., Lengel, R. H. (1986): Organizational Information Requirements, Media Richness and Structural Design, in: Management Science, 32(5), S. 554-571 Davenport, T.H., Jarvenpaa, S.L., Beers, M.C. (1996): Improving Knowledge Work Processes, in: Sloan Management Review, Vol. 37, No. 4, Summer 1996, 53-65 Dennis, A. R., Kinney, S. T. (1998): Testing Media Richness Theory in the New Media - The Effects of Cues, Feedback, and Task Equivocality, in: Information Systems Research, 9(3), S. 256-274. Dennis, R., Valacich, J. (1999): Rethinking Media Richness - Towards a Theory of Media Synchronicity 32nd Hawaii International Conference on Systems Sciences (HICSS 1999), Big Island, Hawaii. DeSanctis, G., Gallupe, R. B. (1987): A Foundation for the Study of Group Decision Support Systems, in: Management Science, Vol. 33, No. 5, 1987, 589-609 Dourish, P. and Belotti, V. (1992): Awareness and Coordination in Shared Workspaces, in: Proceedings of the Conference on Computer-Supported Cooperative Work CSCW'92 (Okt. 31-Nov. 4, Toronto, Canada). ACM, 107-114 Eisenhardt, K. M., Martin, J. A. (2000): Dynamic Capabilities: What Are They?, in: Strategic Management Journal, Vol. 21, 2000, 1105-1121 Ellis, C. A., Gibbs, S. J., & Rein, G. L. (1991): Groupware - Some Issues and Experiences. Communications of the ACM, 34(1), 39-58. Eppler, M.J. (2003): Making Knowledge Visible through Knowledge Maps: Concepts, Elements, Cases, in: Holsapple, C.W. (ed.): Handbook on Knowledge Management, Vol. 1: Knowledge Matters. Springer, Berlin, 189 - 205 Eppler, M.J. (2008): Classifying Knowledge Maps, in: Lytras, M. D., Russ, M., Maier, R., Naeve, A. (eds.): Knowledge Management Strategies: A Handbook of Applied Technologies, Hershey (PA, USA) Eppler, M. J., Reinmann, G. (2007): Wissenswege. Methoden für das persönliche Wissensmanagement. Huber, Bern. Fowler, M. (ed.) (2003): Patterns of Enterprise Application Architecture, Addison Wesley, Boston Frielitz, C., Hippner, H., Wilde, K. D. (2002): eCRM als Erfolgsbasis für Kundenbindung im Internet. in: Bruhn, M., Stauss, B. (eds.): Electronic Services. Diensteistungsmanagement-Jahrbuch 2002, Wiesbaden, 537-562 Garvin, D. A. (1993): Building a Learning Organization, in: Harvard Business Review, Vol. 71, No. 7-8, 1993, 78-91 Gilliland-Swetland, A. J. (2002): Setting the Stage - Introduction to Metadata, http:// www.getty.edu/research/conducting_research/standards/intrometadata, last access 2008-11-01 Göggler, M. (2003): Suchmaschinen im Internet: Funktionsweisen, Ranking-Methoden, Top-Positionen, Springer, Berlin. Grant, R. M. (1991): The Resource-Based Theory of Competitive Advantage: Implications for Strategy Formulation, in: California Management Review, Vol. 33, No. 3, 1991, 114-135

Bibliography

435

Grant, R. M. (1996a): Prospering in Dynamically-Competitive Environments: Organizational Capability as Knowledge Integration, in: Organization Science, Vol. 7, No. 4, July-August 1996, 375-387 Grant, R. M. (1996b): Toward a Knowledge-Based Theory of the Firm, in: Strategic Management Journal, Vol. 17, 1996, Winter Special Issue, 109-122 Gruber, T. R. (1993): A Translation Approach to Portable Ontology Specifications, in: Knowledge Acquisition, Vol. 5, No. 2, 199-220 Gulbins, J., Seyfried, M., Strack-Zimmermann, H. (2002): Dokumentenmanagement, Springer, Berlin Haase, P; Broekstra, J.; Eberhart, A.; Volz, R. (2004): A Comparison of RDF Query Languages, in: McIlraith, S. A.; Plexousakis, D.; van Harmelen, F. (eds.): The Semantic Web - ISWC 2004 Third International Semantic Web Conference, Hiroshima, Japan, November 7-11, 2004, 502 - 518 Hädrich, T. (2008): Situation-oriented Provision of Knowledge Services, PhD Thesis, Martin-Luther-University of Halle-Wittenberg 2008 Hasenkamp, U., Syring, M. (1994): CSCW (Computer Supported Cooperative Work) in Organisationen - Grundlagen und Probleme. In Hasenkamp, U., Kirn, S., Syring, M. (Eds.): CSCW - Computer Supported Cooperative Work. Informationssysteme für dezentralisierte Unternehmensstrukturen (pp. 15-37). Addison Wesley, Bonn. Hall, R. (1992): The Strategic Analysis of Intangible Resources, in: Strategic Management Journal, Vol. 13, No. 2, 1992, 135-144 Heckhausen, H. (1988): Motivation und Handeln. Springer, Berlin Heisig, P., Vorbeck, J., Niebuhr, J. (2001): Intellectual Capital, in: Mertins, K., Heisig, P., Vorbeck, J. (eds.): Knowledge Management. Best Practices in Europe, Berlin et al. 2001, 57-73 IANA (1999): Complete UDP/TCP Port Number List, http://www.iana.org/assignments/portnumbers, last access 2008-11-01 IANA (2001): The Official IANA Registry of MIME-Types, http://www.iana.org/assignments/media-types/, last access 2008-11-01 IEEE (2002): IEEE P1484.1/D11, 2002-11-28 Draft Standard for Learning Technology — Learning Technology Systems Architecture (LTSA), Piscataway (NJ, USA) 2002 Jennex, M., Olfman, L. (2003): Organizational Memory, in: Holsapple, C. W. (ed.): Handbook on Knowledge Management. Vol. 1, Berlin 2003, 207-234 Kaplan, R. S., Norton, D. P. (1996): Using the Balanced Scorecard as a Strategic Management System, in: Harvard Business Review, Vol. 74, No. 1-2, 1996, 75-85 Katz, H (ed.) (2004): XQuery from the Experts - A Guide to the W3C XML Query Language, Addison Wesley, Boston Kirn, S. (2002): Kooperierende intelligente Softwareagenten, Wirtschaftsinformatik 44(1): 53-63 Kohonen, T. (1995): Self-Organizing Maps, 2nd edition, Springer, Berlin Korpela, J. (2004): A Tutorial on Character Code Issues, http://www.cs.tut.fi/~jkorpela/ chars.html, last access 2008-11-01 Krafzig, D., Banke, K., Slama, D. (2005): Enterprise SOA: Service-Oriented Architecture Best Practices, Upper Saddle River 2005

436

Bibliography

Leavitt H.J. (1965): Applied Organisational Change in Industry: Structural, Technological, and Humanistic Approaches, in: March J. (ed.), Handbook of Organisations, Rand McNally & Co. Chicago, 1144-1170 Leonard-Barton, D. (1992): The Factory as a Learning Laboratory, in: Sloan Management Review, Vol. 34, No. 1, 1992, 23-38 Leuf, B., Cunningham, W. (2001): The Wiki Way: Quick Collaboration on the Web, Boston, Addison-Wesley Liebeskind, J. P. (1996): Knowledge, Strategy, and the Theory of the Firm, in: Strategic Management Journal, Vol. 17, 1996, Winter Special Issue, 93-107 Lyman, P., Varian, H.R. (2000): How Much Information, 2000. http://www.sims.berkeley.edu/how-much-info, last access 2008-10-06 Lyman, P., Varian, H.R. (2003): How Much Information 2003, http://www.sims.berkeley.edu/research/projects/how-much-info-2003, last access 2008-10-06 Maier, R. (2007): Knowledge Management Systems. Information and Communication Technologies for Knowledge Management, 3rd edition, Springer, Berlin Maier, R., Remus, U. (2007): Integrating Knowledge Management Services - Strategy and Infrastructure, to appear in: Abou-Zeid, E.-S. (ed.): Knowledge Management and Business Strategies: Theoretical Frameworks and Empirical Research, Hershey (PA, USA) Maier, R., Sametinger, J. (2004): Peer-to-Peer Information Workspaces in Infotop, in: International Journal of Software Engineering and Knowledge Engineering, Vol. 14, No. 1, 79-102 Maier, R., Schmidt, A. (2007): Characterizing Knowledge Maturing. A Conceptual Process Model for Integrating E-Learning and Knowledge Management, in: Proceedings of WM 2007 - 4th Conference on Professional Knowledge Management. Experiences and Visions, Potsdam 2007, 325-333 Mayer, R.E. (2001): Multimedia Learning, Cambridge, Cambridge University Press Mertens, P., Höhl, M. (1999): Wie lernt der Computer den Menschen kennen? Bestandsaufnahme und Experimente zur Benutzermodellierung in der Wirtschaftsinformatik, in: Wirtschaftsinformatik 41 (1999) 3, S. 201-209 Nonaka, I. (1994): A Dynamic Theory of Organizational Knowledge Creation, in: Organization Science, Vol. 5, No. 1, 1994, 14-37 Nonaka, I., Takeuchi, H. (1995): The Knowledge Creating Company, New York 1995 o.V. (2000): What is ... The Speed of (a Definition), http://iroi.seu.edu.cn/books/ee_dic/whatis/thespeed.htm, last access 2008-11-01 O'Dell, C., Grayson, C. J. (1998): If We Only Knew What We Know: Identification and Transfer of Internal Best Practices, in: California Management Review, Vol. 40, No. 3, 1998, 154-174 O’Leary, D. E. (1998): Enterprise Knowledge Management, in: IEEE Computer, Vol. 31, No. 3, 54-61 Pendse, N., Creeth, R. (1995): The OLAP Report. In: Business Intelligence. Porter, M. (1985): Competitive Advantage: Creating and Sustaining Superior Performance, Free Press, New York

Bibliography

437

Porter, M. E. (1980): Competitve Strategy: Techniques for Analyzing Industries and Competitors, New York Porter, M. E. (1990): The Competitive Advantage of Nations, London 1990 Prahalad, C. K., Hamel, G. (1990): The Core Competence of the Corporation, in: Harvard Business Review, Vol. 68, No. 5-6, 1990, 79-91 Priebe, T. (2004): INWISS - Integrative Enterprise Knowledge Portal. Demonstration at the 3rd International Semantic Web Conference (ISWC2004), Hiroshima, Japan, November 2004 Reichwald, R. (1993): Kommunikation und Kommunikationsmodelle, in: Wittmann, W., Kern, W. (Eds.): Enzyklopädie der Betriebswirtschaftslehre - Bd. 1: Handwörterbuch der Betriebswirtschaft, Stuttgart, S. 2174-2188. Reichwald, R. (ed.) (2002): Mobile Komunikation - Wertschöpfung, Technologien, neue Dienste, Gabler, Wiesbaden Röder, S. (2003): Eine Architektur für individualisierte computergestützte Lernumgebungen, Dissertation, Lang, Frankfurt a.M. Ruggles, R. L. (ed.) (1997): Knowledge Management Tools, Boston, MA Saxenian, A. (1994): Regional Advantage: Culture and Competition in Silicon Valley and Route 128. Boston (MA, USA) 1994 Scheer, A.-W. (1994): EDV-orientierte Betriebswirtschaftslehre: Grundlagen für ein effizientes Informationsmanagement, Springer, Berlin Scheer, A.-W. (2001): ARIS - Modellierungsmethoden, Metamodelle, Anwendungen, Springer, BerlinSchmidt, A. (2005): Knowledge Maturing and the Continuity of Context as a Unifying Concept for Integrating Knowledge Management and E-Learning, in: Proceedings of I-KNOW 05 - 5th International Conference on Knowledge Management, Special Track on Integrating Working and Learning, Graz, June 29th-July 1st, 2005 Schulz von Thun, Friedemann (1981): Miteinander reden. 1. Störungen und Klärungen. Allgemeine Psychologie der Kommunikation. Rowohlt Taschenbuch Verlag, Reinbeck. Schultze, U. (2000): A Confessional Account of an Ethnography about Knowledge Work, in: MIS Quarterly, Vol. 24, No. 1, 2000, 1-39 Searle, J.R. (1969): Speech Act: An Essay in the Philosophy of Language, Cambridge, Cambridge University Press Shannon, C. E., & Weaver, W. (1949): The Mathematical Theory of Communication. Urbana IL, University of Illinois Press Siegel, Jon (2000): CORBA 3 Fundamentals and Programming, 2nd edition, Wiley, New York Skyrme, D., Amidon, D. (1997): The Knowledge Agenda, in: The Journal of Knowledge Management, Vol. 1, No. 1, 1997, 27-37 Spender, J.-C. (1994): Organizational Knowledge, Collective Practice and Penrose Rents, in: International Business Review, Vol. 3, No. 4, 1994, 353-367 Spender, J.-C. (1996): Organizational Knowledge, Learning and Memory: Three Concepts in Search of a Theory, in: Journal of Organizational Change Management, Vol. 9, No. 1, 1996, 63-78

438

Bibliography

Spender, J.-C. (1996a): Making Knowledge the Basis of a Dynamic Theory of the Firm, in: Strategic Management Journal, Vol. 17, 1996, Winter Special Issue, 45-62 Staab, S., Studer, R., Schnurr, H.-P., Sure, Y. (2001): Knowledge Processes and Ontologies, in: IEEE Intelligent Systems & their Applications, Vol. 16, No. 1, 26-34 Stein, E., Zwass, V. (1995): Actualizing Organizational Memory with Information Systems, in: Information Systems Research, Vol. 6, No. 2, 1995, 85-117 Stewart, T. A. (1997): Intellectual Capital: The New Wealth of Organisations, New York 1997 Suresh Raj, G. (2002): The Advanced COM/DNA Tutorial, http://my.execpc.com/~gopalan/ com/com_tutorial.html, last access 2008-11-01 Sveiby, K.-E. (1997): The New Organizational Wealth. Managing and Measuring Knowledge-Based Assets, San Francisco 1997 Sveiby, K.-E., Lloyd, T. (1987): Managing Knowhow, London; based on: Sveiby, K.-E., Risling, A.: Kunskapsföretaget (in Swedish; the Know-How Organization), Malmö 1986, also published in German as Sveiby, K.-E., Lloyd, T.: Das Management des Know-How. Führung von Beratungs-, Kreativ- und Wissensunternehmen, Frankfurt/ New York 1990 Sweller, J. (1999): Instructional Design in Technical Areas, Camberwell, ACER Press Szyperski, C. (1997), Component Software: Beyond Object-Oriented Programming, Addison-Wesley Turowski, K, Pousttchi, K. (2004): Mobile Commerce - Grundlagen und Techniken, Springer, Berlin Watson, R. T. (2005): Data Management. Data Bases & Organizations, 5th edition, Wiley, New York Watson, R. T. (2008): Evolutionary Information Systems, Presentation at the University of Innsbruck, 2008 WFMC (1995): The Workflow Reference Model, http://www.wfmc.org/, last access 200811-01 WFMC (2002): An Introduction to Workflows, Section from Workflow Handbook 2002, http://www.wfmc.org/, last access 2008-11-01 Wiig, K. M. (1997): Integrating Intellectual Capital and Knowledge Management, in: Long Range Planning, Vol. 30, No. 3, 1997, 399-405 Winograd, T., Flores, F. (1986): Understanding Computers and Cognition: A New Foundation for Design, Ablex Publ. Corp., Norwood Wooldridge, M., Jennings, N.R. (1995): Intelligent Agents - Theory and Practice, in: The Knowledge Engineering Review 10(2) Zack, M. H. (1999): Managing Codified Knowledge, in: Sloan Management Review, Vol. 40, No. 4, Summer 1999, 45-58 Zigurs, I., Buckland, B. K. (1998): A Theory of Task/Technology Fit and Group Support Systems Effectiveness, in: Management Information Systems Quarterly - MISQ, Vol. 22, No. 3, 1998, 313-334 Zschau,O., Traub, D., Zahradka, R. (2002): Websites professionell planen und betreiben, 2nd edition, Galileao Press, Bonn

Index

Symbols .NET 143, 381 Numerics 2-tier architecture 48 3-tier architecture 48 A access control list 193 access point 113 access privilege 318 access privileges 318 access service 78 account 192 account management 194 accounting system 58 ACL 193 ActiveX 378 address book 292 ad-hoc learning 40 ad-hoc workflow 228 adhocracy 9 after-action review 38 agent architecture 53 agile 71 AJAX 375 AML 203 animations 345 antivirus program 137 application programming interface 219 application server 138 application sharing 299 architecture 45, 75 archiving 326 ARP 102 artificial intelligence 29 asset 308 ATM 96 audio conference 298 audit 321 authentication 134, 192 authoring 349 authoring tool 347 authorization 192 awareness 293 axis 167

B backup 131 Balanced Scorecard 15 barcode 317 Basic Multilingual Plane 157 behaviorism 334 best practice 39, 41 blended learning 339 blog 277 blue pages 37 Bluetooth 93 body area network 88 boolean operators 261 BPEL 203 BPML 203 BPMN 203 bridge 112 broadcast 101 broadcast network 85 bus network 86 business information system 56 business process execution language 232 business system layer 202 C calendar 292 capability differential 12 capturing 312, 314 cascading stylesheet 165, 375 case debriefing 39 case-based reasoning 39 certificate 44, 132 certification authority 132 chat 297 check-out 321 choreography 232 classification of computer networks 84 click stream analysis 372 code of practice 330 coded information 314 codification 66 cognitive architecture 54 cognitivism 334 COLD 327 collaborative authoring 301 collaborative filtering 373 collaborative tagging 275 collection 182 combination 19 commitment 45 common-sense ontology 209 communication model 286 community management 42 competence 272, 278, 282 competence management 37 compliance 328

440

Index

component, software 138 component-oriented programming 138 compression 260 computer integrated manufacturing 60 computer-based training 339 conceptual architecture 46 connectivity 395 constructivism 334 container 181 content 307, 308 content meta-data 173 content-based filtering 373 context 7, 20, 45, 167, 283, 288, 414 context meta-data 173 contextualized knowledge 19 controller, storage 118 conversation network 289 CORBA 141, 222 core competency 12 core process 59 crawler 258 customer relationship management 60 customization 373 D DAML+OIL 214 data administration 31 data constraint language 186 data definition language 186 data management 31 data manipulation language 186 data protection law 329 data source logic 49 data work 23 database administration 30 database management system 120 DCL 186 DDL 186 declarative knowledge 20 DECT 96 definition of business process 226 definition of digital identity 194, 207 definition of enterprise knowledge infrastructure 69 definition of KM instrument 34 definition of knowledge 16, 23 definition of knowledge management 32 definition of service 70 definition of software agent 54 definition of WfMS 228 definition of workflow 227 deliberative architecture 54 depreciation of knowledge 12

desktop alternative 383 desktop metaphor 383 desktop search engines 263 DHCP 110 digital signature 132 direct attached storage 124 directory service 120 distance learning 338 DML 186 DNS 111, 193 document 309 document classification 317 document instance 159 document management 309 document management system 310 document object model 376 document type declaration 159 domain logic 49 domain ontology 208 drive, storage 117 driver, storage 118 Dublin Core 154, 176 E EAI 70 Eclipse 381 ECM 310, 360 EDGE 98 e-learning 338 e-learning platform 355 e-learning standard 350 electronic signature 329 email 108, 300 encryption 131 encryption, asymmetric 132 encryption, symmetric 132 enterprise content management 310 Enterprise Java Beans 142 enterprise resource planning 60 enterprise service bus 145 entity-relationship model 153 EPK 203 Ethernet 94 evaluation 359 event-driven process chain 203 executive information system 58 experience 272, 273 experience management 273 experience management system 37, 278 expert 283 expert advice 40 expertise 14, 273, 281 expertise locator 37, 284

Index explicit knowledge 19 expression 272 eXtensible Markup Language 157 eXtensible Stylesheet Language Transformation 165 externalization 19 Extranet 128 F FDDI 96 feature phone 396 federated search 263 file server 122 filesystem 118 firewall 136 Firewire 92 Flash 379 FLWOR 188 folksonomy 275 Frame relay 96 frequently asked questions 284 FTP 106 function 219 fuzzy logic 261 G gateway 113 geo coordinate 199 geo data 196 geo shape 198 GeoRSS 198 GML 198 good practice 41 Google 51 GPRS 98 GPS 197 group awareness 293 group decision support system 303 Groupware 302, 384 Groupware client 390 GSM 97 H hierarchical storage management 122 history of e-learning 338 HSCSD 98 HTML 5 377 HTTP 105 HTTPS 134 hub 112 human-oriented KM 32 hybrid architecture 54 hyperbolic tree 255

hypermedia systems 339 hypertext organization 202 I ICMP 103 ID3 175 IDE 118 identification 192 identity management 194 IMAP 108 index 260 indexing 260 individuation 44, 272 in-form-ation 44 information agent 55 information hardness 45 information life-cycle management 122 information management 31 information resource 307 information resource life-cycle 306 information resource management 31 information work 23 informing practice 43 Infotop 255 infrastructure service 76 instant messaging 297 instruction 44 intangible asset 10 Intangible Assets Monitor 15 integration service 77 intellectual capital 15 Intellectual Capital Navigator 15 intelligent character recognition 316 interaction 44 interface agent 56 internalization 19 Internet 126 Internet protocol 102 Internet search engines 262 InterRAP 54 Intranet 127 Intranet search engines 262 intrusion detection systems, IDS 136 intuitive 283 investigation 44 IP address 100 IPsec 133 IrDA 92 IS pyramid 56 ISO/OSI reference model 89 J J2EE 141

441

442

Index

Java 380 Java applet 378 Java FX 378 Java RMI 221 JDOQL 187 JSF 377 just-in-time KM 74 K Kerberos 135 KM initiative 34, 66 KM instrument 34, 64 KM measure 34 KM project 34 KM service 71 KM service infrastructure 71 KML 196, 198 knowledge 4 knowledge application map 38 knowledge as product 8 knowledge asset map 38, 280 knowledge base layer 203 knowledge development map 38 knowledge economy 1 knowledge element 20 knowledge management 31 knowledge management system 66 knowledge map 38 knowledge maturing 43 knowledge network 42 knowledge process 8, 77 knowledge process reengineering 42 knowledge repository 151 knowledge routine 40 knowledge service 71, 77 knowledge society 1 knowledge source map 38 knowledge structure map 38 knowledge topic 65 knowledge warehouse 66 knowledge work 22 knowledge worker 22 knowledge-based view 10 L LDAP 122, 192 learning content management system 66 learning goal 344 learning goals 344 learning management platform 66 learning management portal 66 learning management system 66, 354 learning object 341

learning object repository 66 learning organization 28 learning unit 343 legitimation 45 lessons learned 38 lexical analysis 258 LINQ 189 list server 300 load balancing 139 local area network 88 localhost 102 location-based service 391 logical layer 218 loop network 87 M MAC-address 100 machine bureaucracy 9 management information system 58 management work 23 mashup 196 master data 192 maxim 283 media richness 295 media synchronicity 295 medium 17 medium, network 91 medium, storage 115 meeting support system 299 mesh network 87 message queue 145 message-oriented middleware 145 meta-data dimension 191 meta-data ontology 208 method 219 method and task ontology 209 metropolitan area network 88 microarticle 37 microblog 278 microformats 184 MIME 108 mindmapping 274 mobile computing 390 mobile device 394 mobile framework 400 mobile operating system 400 mobile phone 396 modem 114 MoReq2 331 motivation 335 MPEG-7 154, 176 multimedia applications 385 multimedia design principles 349

Index multi-user editor 301 N namespace 160 net-based locating 391 network address translation, NAT 136 network attached storage 125 network class 100 network effect 426 network interface card 113 networking 292 neural networks 265 news aggregation 268 newsgroup 301 NFS 107 NNTP 107 node test 167 non-coded information 314 notebook 398 notification 270 novice 282 O object oriented programming 138 ODMA 311 office information system 58 Office Open XML 385 offline reading 274 on-demand 71 on-demand KM 74 ontology 65 ontology mapping 215 OODBMS 185 OpenDocument 385 operational system 56 optical character recognition 316 OQL 185, 187 orchestration 232 organizational asset 10 organizational learning 28 organizational memory 29 organizational memory system 66 organizational proposal system 37 organizational resource 10 OWL 213 OWL-S 235 P participant 67 partition 118 pattern matching 261 PDF/A 315 peer-to-peer architecture 51, 421

443

peer-to-peer architecture with super peers 53 peer-to-peer architecture, assisted 53 peer-to-peer architecture, pure 53 persistence 139 personal area network 88 personal digital assistant 396 personal experience management 37 personal information agent 270 personal knowledge management 280 personal learning environment 281 personalization 66, 195, 284, 367, 370 PGP 137 phone 298 physical architecture 46 plug-in technology 387 podcast 268 point-to-point network 86 POP3 108 portal 366 portal server 369 portlet 369 predicate 167 presentation logic 49 primary activity 58 privilege 192 procedure 219 process landscape 203 processing instruction 158 process-oriented KM 72 professional bureaucracy 9 profiling 370 programmed instruction 339 project debriefing 38 project team layer 202 proof layer 218 property element 180 protection of knowledge 7 proxy 136 public key infrastructure 133 Q qualified name 161 quantity of knowledge 8 query analysis 260 R RAID 123 ranking 265 raster data 197 RDF graph 179 RDF instance 213 RDF query language 189 RDF Schema 210

444

Index

RDF statement 179 RDF triple 179 RDF/XML 184 RDFQL 190 RDQL 190 reactive architecture 54 recommendation services 256 record 119 records management 310 regular expressions 262 reification 181 relational DBMS 186 relational theory 153 relevance feedback 267 remote access service, RAS 135 remote GUI 379 remote procedure call 220 repeater 112 representational ontology 209 Resource Description Framework 178 resource-based view 9 rich client 49, 380 rich Internet applications 378 rich thin client 378 ring network 86 role 192 roundtrip 374 router 113 routing 102 RSS 198, 268 rules layer 218 S S/MIME 137 sandbox 378 SAP NetWeaver 63 SAP Netweaver 203 SAP R/3 63 scanning 313 SCSI 118 security group 192, 193 security threats 129 semantic content management 38 semantic search engines 264 Semantic Web initiative 154 Semantic Web services 234 Semantic Web stack 206 sequencing 349 serialization 179 SeRQL 190 server services 50 service 70, 219 service work 23

service-oriented architecture 70 services sciences 70 shared information spaces 303 Silverlight 379 simulation 345 single instance storage 327 single sign-on 367 Skandia Navigator 15 skill management 37 smartphone 396 smarttag 389 SMB 107 SMS 109 SMTP 108 SNMP 110 SOA 70 SOAP 70, 223 social bookmarking 275 social context 17 social software 290 social system 20 socialization 19 SONET/SDH 96 spam filter 137 SPARQL 190 speech act 288 SQL 186 SSL 134 staging 322 star network 87 stemming 259 stereotype 371 stop words 259 storage area network 124 strategic capability 12 structure meta-data 173 Stud.IP 356 subject matter expert 65 subnotebook 397 supplier relationship management 60 supply chain management 60 support activity 59 supporting process 59 switch 113 synchronization 401 SyncML 402 syndication 268 T tacit dimension 19 tacit knowledge 19 tag 274 tagging 274

Index tangible asset 10 taxonomy 207 TCP 103 technology-enhanced learning 41, 338 technology-oriented KM 32 telnet 112 template 322 templates 322 terminal 50 terminal-based locating 392 thin client 49, 374 Tobin’s q 15 Token Ring 94 traditional Web publishing 323 transaction 233 transaction data 192 transaction handling 139 transaction processing system 56 transfer of knowledge 6 tree network 87 trust layer 218 types of knowledge 17 types of learners 335 types of ontologies 208 U UCS 360 UDDI 70, 223 UDP 103 UMTS 98 Unicode 157 unified communication server 360 unified communications 109 unified messaging 299 uniform resource identifier 156 uniform resource locator 156 uniform resource name 156 USB 92 user 192 user account 192 user model 370, 371 user tracking 372 V validity 45 value chain 58, 203 vCard 292 vector data 197 version 320 video conference 298 virtual private network, VPN 134 virtual storage 326 voice telephony 395

VoIP 109 volume 118 VPN 134 W WAP 107, 401 Web 2.0 424 Web content management system 321 Web mining 373 Web Ontology Language 213 Web service 70 Web service composition 234 Web service discovery 236 Web services 222 Web-based training 339 WebDAV 107 Webtop 427 well-formedness 160 wide area network 88 Wi-Fi protected access 133 Wiki 302 wiki 302 wired equivalent privacy, WEP 133 Wireless LAN 95 wireless media 91 worflow engine 231 workflow 227 workflow life-cycle 228 workflow management 227 workplace learning 74 WPA 133 WSDL 222 WSMO 235 X X.25 96 XAML 376 XHTML 184, 185 XML 70, 157 XML database 185 XML declaration 159 XML-messaging 172 XPath 166, 188 XPDL 203 XQuery 188 XSLT 165 XSLT engine 172 Y yellow pages 37

445