Building the e-World Ecosystem: 11th IFIP WG 6.11 Conference on e-Business, e-Services, and e-Society, I3E 2011, Kaunas, Lithuania, October 12-14, 2011, Revised Selected Papers

IFIP Advances in Information and Communication Technology

353

Editor-in-Chief A. Joe Turner, Seneca, SC, USA

Editorial Board Foundations of Computer Science Mike Hinchey, Lero, Limerick, Ireland Software: Theory and Practice Bertrand Meyer, ETH Zurich, Switzerland Education Arthur Tatnall, Victoria University, Melbourne, Australia Information Technology Applications Ronald Waxman, EDA Standards Consulting, Beachwood, OH, USA Communication Systems Guy Leduc, Université de Liège, Belgium System Modeling and Optimization Jacques Henry, Université de Bordeaux, France Information Systems Jan Pries-Heje, Roskilde University, Denmark Relationship between Computers and Society Jackie Phahlamohlaka, CSIR, Pretoria, South Africa Computer Systems Technology Paolo Prinetto, Politecnico di Torino, Italy Security and Privacy Protection in Information Processing Systems Kai Rannenberg, Goethe University Frankfurt, Germany Artificial Intelligence Tharam Dillon, Curtin University, Bentley, Australia Human-Computer Interaction Annelise Mark Pejtersen, Center of Cognitive Systems Engineering, Denmark Entertainment Computing Ryohei Nakatsu, National University of Singapore

IFIP – The International Federation for Information Processing IFIP was founded in 1960 under the auspices of UNESCO, following the First World Computer Congress held in Paris the previous year. An umbrella organization for societies working in information processing, IFIP’s aim is two-fold: to support information processing within ist member countries and to encourage technology transfer to developing nations. As ist mission statement clearly states, IFIP’s mission is to be the leading, truly international, apolitical organization which encourages and assists in the development, exploitation and application of information technology for the benefit of all people. IFIP is a non-profitmaking organization, run almost solely by 2500 volunteers. It operates through a number of technical committees, which organize events and publications. IFIP’s events range from an international congress to local seminars, but the most important are: • The IFIP World Computer Congress, held every second year; • Open conferences; • Working conferences. The flagship event is the IFIP World Computer Congress, at which both invited and contributed papers are presented. Contributed papers are rigorously refereed and the rejection rate is high. As with the Congress, participation in the open conferences is open to all and papers may be invited or submitted. Again, submitted papers are stringently refereed. The working conferences are structured differently. They are usually run by a working group and attendance is small and by invitation only. Their purpose is to create an atmosphere conducive to innovation and development. Refereeing is less rigorous and papers are subjected to extensive group discussion. Publications arising from IFIP events vary. The papers presented at the IFIP World Computer Congress and at open conferences are published as conference proceedings, while the results of the working conferences are often published as collections of selected and edited papers. Any national society whose primary activity is in information may apply to become a full member of IFIP, although full membership is restricted to one society per country. Full members are entitled to vote at the annual General Assembly, National societies preferring a less committed involvement may apply for associate or corresponding membership. Associate members enjoy the same benefits as full members, but without voting rights. Corresponding members are not represented in IFIP bodies. Affiliated membership is open to non-national societies, and individual and honorary membership schemes are also offered.

Tomas Skersys Rimantas Butleris Lina Nemuraite Reima Suomi (Eds.)

Building the e-World Ecosystem 11th IFIP WG 6.11 Conference on e-Business, e-Services, and e-Society, I3E 2011 Kaunas, Lithuania, October 12-14, 2011 Revised Selected Papers

13

Volume Editors Tomas Skersys Rimantas Butleris Lina Nemuraite Kaunas University of Technology Studentu g. 50-313a, 51368 Kaunas, Lithuania E-mail: {tomas.skersys, rimantas.butleris, lina.nemuraite}@ktu.lt Reima Suomi University of Turku Turku School of Economics Rehtorinpellonkatu 3, 20500 Turku, Finland E-mail: [email protected]

ISSN 1868-4238 e-ISSN 1868-422X e-ISBN 978-3-642-27260-8 ISBN 978-3-642-27259-2 DOI 10.1007/978-3-642-27260-8 Springer Heidelberg Dordrecht London New York Library of Congress Control Number: 2011943175 CR Subject Classification (1998): J.1, C.2, H.4, H.3, K.6.5, D.2

© International Federation for Information Processing 2011 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, re-use of illustrations, recitation, broadcasting, reproduction on microfilms 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 ist 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. Typesetting: Camera-ready by author, data conversion by Scientific Publishing Services, Chennai, India Printed on acid-free paper Springer is part of Springer Science+Business Media (www.springer.com)

Preface

The I3E conference series is now well in its second decade, and you are holding in your hands the 11th edition of I3E proceedings. Issues around e-society seem to become increasingly more important every day. At the same time, the number of conferences, journals and other forums dedicating room to these issues is growing dramatically. The I3E conference has kept its position as one of the leading academic conferences dealing with e-World issues. As a flagship conference of the IFIP WG6.11 “Communication Aspects of the E-World,” the conference has gained a lot of attention and approval from leading academics in the field, which is reflected in the quality of its proceedings. As before, this year the conference articles address a broad scope of e-World topics ranging from basic technologies such as service-oriented architecture or semantic ontologies to many types of applications, such as education systems, e-health solutions or e-government. As ever, this publication is an outcome of the work of many dedicated stakeholders. We would like to thank our basic supporting organizations, IFIP and Springer, for making this proceedings book possible. The conference would not have been possible without the great support of Kaunas University of Technology. The biggest thank-you of course goes to the authors, who decided to share their ideas and knowledge presented in the papers. On the other hand, one cannot forget the editing process behind which there are technical editors, a number of brilliant reviewers as well as local organizers of the event. We would like to thank all of them deeply. The I3E conference series looks eagerly to the further challenges of the second decade of its operations. Many places have already shown their interest in organizing future conferences. Despite the fast-paced nature of modern science, there is no doubt research and activity on e-World issues will continue with increasing intensity for some time to come. Our task is to catch some of this phenomenon under the lens of high-quality science and document it. We have every reason to believe the knowledge gained will contribute to a better future e-World. October 2011

Rimantas Butleris Rita Butkiene Reima Suomi Lina Nemuraite Tomas Skersys

Organization

The 11th IFIP Conference on e-Business, e-Services, e-Society (I3E 2001) was organized by Kaunas University of Technology and took place in Kaunas, Lithuania (October 12–14, 2011). The conference was sponsored by IFIP WG 6.11 in cooperation with TC11 and TC8.

General Chair Rimantas Butleris

Kaunas University of Technology, Lithuania

Steering Committee Wojciech Cellary Winfried Lamersdorf Reima Suomi

Poznan University of Economics, Poland University of Hamburg, Germany University of Turku, Finland

Local Organizing Committee Rita Butkiene (Chair) Tomas Danikauskas Kestutis Kapocius Edvinas Sinkevicius Tomas Skersys Lina Tutkute

Kaunas Kaunas Kaunas Kaunas Kaunas Kaunas

University University University University University University

of of of of of of

Technology, Technology, Technology, Technology, Technology, Technology,

Lithuania Lithuania Lithuania Lithuania Lithuania Lithuania

Program Committee Lina Nemuraite (Chair) Reima Suomi (Chair) Americo Nobre Amorim Eduardas Bareisa Melanie Bicking Wojciech Cellary Kok-Wai Chew Dirk Deschoolmeester Irene Krebs Winfried Lamersdorf

Kaunas University of Technology, Lithuania University of Turku, Finland Ufpe/FIR, Brazil Kaunas University of Technology, Lithuania Universit¨ at Koblenz-Landau, Germany Poznan University of Economics, Poland Multimedia University, Malaysia University of Ghent, Belgium University of Technology, Germany Universit¨ at Hamburg, Germany

VIII

Organization

Hongxiu Li Zoran Milosevic Yuko Murayama Harri Oinas-Kukkonen Michael Pantazoglou Vytautas Pilkauskas Riedl Reinhard Dalius Rubliauskas Santosh Shrivastava Roland Traunm¨ uller Weijun (Joe) Wang Hans Weigand Jing Zhao Hans-Dieter Zimmermann

Turku School of Economics, Finland Deontik, Australia Iwate Prefectural University, Japan University of Oulu, Finland NKUA, Greece Kaunas University of Technology, Lithuania Applied University of Bern, Switzerland Kaunas University of Technology, Lithuania Newcastle University, UK Universit¨ at Linz, Austria Central China Normal University, China Tilburg University, The Netherlands China University of Geosciences, China IPM der FHS St. Gallen, Switzerland

Additional Reviewers Arunas Andziulis Tomas Blazauskas Germanas Budnikas Lina Ceponiene Audrone Janaviciute Vacius Jusas Dalia Krisciuniene Regina Miseviciene Nerijus Morkevicius Algirdas Noreika Vytautas Pilkauskas Rimantas Plestys Tomas Skersys Bronius Tamulynas

Klaipeda University, Lithuania Kaunas University of Technology, Kaunas University of Technology, Kaunas University of Technology, Kaunas University of Technology, Kaunas University of Technology, Vilnius University, Lithuania Kaunas University of Technology, Kaunas University of Technology, Kaunas University of Technology, Kaunas University of Technology, Kaunas University of Technology, Kaunas University of Technology, Kaunas University of Technology,

Lithuania Lithuania Lithuania Lithuania Lithuania Lithuania Lithuania Lithuania Lithuania Lithuania Lithuania Lithuania

Co-editors Tomas Skersys Rimantas Butleris Lina Nemuraite Reima Suomi

Kaunas University of Technology, Lithuania Kaunas University of Technology, Lithuania Kaunas University of Technology, Lithuania University of Turku, Finland

Table of Contents

e-Government and e-Governance A Practical Methodology for Development of a Network of e-Government Domain Ontologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hele-Mai Haav

1

Channel Choice in the G2B Context: Towards a Research Agenda . . . . . . Yvon van den Boer, Willem Pieterson, and Rex Arendsen

14

Supporting Customs Controls by Means of Service-Oriented Auditing . . . Hans Weigand and Faiza Allah Bukhsh

28

IMP2.0 Migration Collaborative Platform and e-Services . . . . . . . . . . . . . . Nineta Polemi, George Pentafronimos, and Theodoros Ntouskas

44

e-Services Consumer Side Resource Accounting in the Cloud . . . . . . . . . . . . . . . . . . . . Ahmed Mihoob, Carlos Molina–Jimenez, and Santosh Shrivastava

58

Privacy Aware Obfuscation Middleware for Mobile Jukebox Recommender Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ahmed M. Elmisery and Dmitri Botvich

73

Self-service Systems Performance Evaluation and Improvement Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Darius Dilijonas and Virgilijus Sakalauskas

87

Digital Goods and Products Service Provisioning Framework for Digital Smart Home Services . . . . . . Egidijus Kazanavicius, Darius Imbrasas, and Mantas Razukas

99

Open-Source Software Implications in the Competitive Mobile Platforms Market . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Salman Qayyum Mian, Jose Teixeira, and Eija Koskivaara

110

Diminishing Print vs. Rising Electronic Publishing: The Case of Lithuania . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gintare Gudaityte, Vida Melninkaite, Daiva Vitkute-Adzgauskiene, and Vladislav V. Fomin

129

X

Table of Contents

e-Business Process Modeling and Re-engineering A Method for Conceptual Modeling of Semantically Integrated Use-Case Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Remigijus Gustas and Prima Gustiene

138

Ontology-Driven Business Process Design . . . . . . . . . . . . . . . . . . . . . . . . . . . Carla Marques Pereira, Artur Caetano, and Pedro Sousa

153

On the Use of a Role Ontology to Consistently Design Business Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Artur Caetano and Jos´e Tribolet

163

Towards Traceability Metamodel for Business Process Modeling Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Saulius Pavalkis, Lina Nemuraite, and Edita Mileviˇcien˙e

177

Innovative e-Business Models and Implementation The Connected Home: From Market Barriers to Business Model Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Malgorzata Ciesielska and Feng Li

189

Developing Sustainable e-Business Models for Intelligent Transportation Systems (ITS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Konstantinos N. Giannoutakis and Feng Li

200

Observations on e-Business Implementation Capabilities in Heterogeneous Business Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tero Pesonen and Kari Smolander

212

Development of e-Services in Latvian Insurance Industry: The Critical Role of Quality Benchmarking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Valdis Janovs and Vladislav V. Fomin

227

e-Health and e-Education A New Model to Study ICT Adoption in Connection with Physical Activity – The TAMPA-Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Eija Koskivaara, Raija Laukkanen, and Olli J. Heinonen

237

A New Electronic Path to the Hospital – A Prenatal Observation System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Arnim Nethe and Irene Krebs

245

Students’ e-Learning System Usage Outcomes: A Study with a Learning Management System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A.K.M. Najmul Islam

255

Table of Contents

Antecedents of Post-adoption Expectations: A Study in the e-Learning Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A.K.M. Najmul Islam

XI

269

Innovative e-Business Models Towards Reuse-Oriented and Web-Based Collaborative Framework for e-Business Providers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vladimirs Kotovs, Andrejs Lesovskis, and Leonids Novickis

282

Weibull Model for Dynamic Pricing in e-Business . . . . . . . . . . . . . . . . . . . . Nicholas Nechval, Maris Purgailis, and Konstantin Nechval

292

Multi-level Fuzzy Rules-Based Analysis of Virtual Team Performance . . . Dalia Kriksciuniene and Sandra Strigunaite

305

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

319

A Practical Methodology for Development of a Network of e-Government Domain Ontologies Hele-Mai Haav Institute of Cybernetics at Tallinn University of Technology, Akadeemia 21, 12618 Tallinn, Estonia [email protected]

Abstract. Ontologies are widely proposed for solving semantic interoperability problems in e-government domain. However, less attention is paid on methodological issues of ontology development in this domain. These are not sufficiently solved. Existing approaches in ontology development are technically very demanding and require long learning curve and are therefore inefficient for this purpose. E-government domain ontologies could be efficiently developed by domain experts rather than ontology engineers. The paper proposes a novel ontology development methodology for construction of light-weight domain ontologies by bringing ontology modeling closer to domain experts that are actual domain knowledge holders. The approach is illustrated and evaluated by development of a network of e-government domain ontologies used for achieving semantic interoperability of state information systems (IS) in Estonia. Keywords: ontology, ontology networks, ontology development methodology, e-government.

1

Introduction

Applicability of semantic technologies in the e-Government domain has been investigated for many years. A number of EU projects (e.g. SemanticGov [11], SmartGov [3], OntoGov [14], Access-eGOV [2], etc) are devoted to the several aspects of using ontologies in e-government domain and for different applications. Most of the projects are dealing with building semantic e-government services (i.e. electronic governmental services enhanced by semantic annotations) in order to provide their semantic interoperability. Many member states of the European Union have started some semantic interoperability initiatives as for example Italian’s initiative in public administration1, Finnish semantic initiative FinnONTO [7], Semantic Latvia project [1] and Estonian Semantic Interoperability Framework [6, 15]. There are also pan-European initiatives, which include SEMIC (SEMantic Interoperability Centre Europe)2 and SemanticGov project [11]. 1 2

http://www.digitpa.gov.it http://semic.eu/about_semantic_interoperability_centre_europe.html

T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 1–13, 2011. © IFIP International Federation for Information Processing 2011

2

H.-M. Haav

However, practical effects of this research and initiatives to e-government domain and applications do not well meet expectations. One of the main reasons as three years experience of Estonian semantic interoperability initiative shows is complexity of the semantic modeling (e.g. ontology development) and annotation process. There is lack of ontology development methodologies that are not technically demanding in order to be used by e-government domain experts and that are easy to learn by employees of public administrations (PA-s). This paper concentrates to the problem of complexity of semantic modeling by answering the question: How to easily develop domain ontologies that could be used for semantic annotation of e-government data services? Majority of current ontology development methodologies [4, 5, 13, 16] require a role of mediator (e.g. knowledge engineer) between actual knowledge holders (e.g. domain experts) and ontology engineers who encode knowledge in one of available onotology formalisms (e.g. OWL3 ). In contrast, this paper proposes a novel ontology development methodology for construction of light-weight domain ontologies by bringing ontology modeling closer to domain experts (PA employees) that are actual knowledge holders. The methodology could be used for construction of domain ontologies by domain experts without any high level knowledge of semantic technologies. The approach is illustrated and evaluated by development of network of egovernment domain ontologies used for annotation of governmental data services in order to support semantic interoperability of state information systems (IS) in Estonia. The paper is structured as follows. Section 2 gives a brief overview of related semantic e-government projects and ontology creation methodologies. In section 3 a motivating Estonian e-government case study is described and analysed. Section 4 presents a practical ontology development methodology for network of e-government domain ontologies. Section 5 is evaluation of the methodology and section 6 concludes the work.

2

Related Works

Majority of e-government projects use ontologies for achieving semantic interoperability and integration between e-Government systems. For example, in OntoGov [14] ontologies are constructed for representing the meaning of processed data, resources and provided functionality of e-Government services. However, it is not always clear, what methodology is used for the ontology development in egovernment projects. Usually top-down approach is used. There are some general methodological approaches (e.g., METHONTOLOGY [4], On-To-Knowledge [13], etc.) for ontology construction. More detailed overview of wide range of methodologies can be found in [4].

3

OWL DL, www.w3.org/TR/owl-guide

A Practical Methodology for Development of a Network

3

The most well-know methodology is METHONTOLOGY methodology [4] that enables the construction of ontologies at the knowledge level. METHONTOLOGY is based on the main activities identified by the software development process [8] and by knowledge engineering methodologies. This methodology includes the ontology development process, a life cycle based on evolving prototypes and techniques to carry out each activity in the management, development and support activities. Another approach is TOVE [16] where authors suggest using questionnaires, which is useful where domain experts have very little knowledge of knowledge modeling. There are also attempts to improve existing methodologies by incorporating Rapid Application Development (RAD) approaches in ontology development [10, 17]. The NeOn [5] ontology development methodology takes software engineering approach and goes step further having an ambitious goal to serve as a methodology for building a very large number of ontologies embedded in ontology networks built collaboratively by distributed teams. NeOn methodology is also supported by several tools collected into the NeOn toolkit [5]. All these approaches have turned ontology construction to an engineering activity. However, there is lack of methodologies that follow a user-oriented (e.g. domain expert oriented) approach. The majority of approaches require extensive technical knowledge of formal languages and techniques for capturing knowledge of a domain. Therefore, this paper is devoted to bringing ontology modeling closer to domain experts. The proposed practical methodology for e-government ontology building is taking into account some of the proposals from the NeOn methodology as egovernment ontologies in principle are networked ontologies.

3

Use Case: Semantic Interoperability of e-Government Services in Estonia

In Estonia, interoperability of state IS-s is achieved by using middleware platform XRoad [9] that provides the technical interoperability of the components and registries of state IS. X-Road is a technical and organizational environment that enables secure data transfer between digital government databases and enables secure data transfer between individuals and government institutions. X-road platform is based on SOAP4 standard and governmental e-service descriptions are presented in WSDL5 language. According to the statistics of the using X-Road [9] the estimated number of uses in 2010 was 100 million. The X-Road does not support semantic interoperability. In order to initiate semantic interoperability activities, Ministry of Economic Affairs issued strategic document on semantic interoperability of state information technology (IT) in 2005. This document presents a vision of semantic interoperability of state information systems in Estonia. Following this vision, the semantic interoperability architecture for Estonian state IS-s was proposed in 2007 [6]. It is designed as the semantic layer to be built on top of X-Road infrastructure. 4 5

SOAP standard, http://www.w3.org/TR/soap12-part1/#intro WSDL, http://www.w3.org/TR/wsdl

4

H.-M. Haav

Important components of semantic interoperability architecture are domain ontologies of state IS-s. Primary use of ontologies as foreseen by the architecture is semantic annotation of descriptions of e-government data services and data structures of databases of state IS-s. As WSDL descriptions of existing services are available then SAWSDL6 has been chosen for enabling these to be enhanced with semantic descriptions. For data structures, their XMI7 descriptions are enriched with SAWSDL references to corresponding components of domain ontologies. As a repository of ontologies and semantic descriptions, the Administrative System of the State Information System (RIHA) [12] is used. It is a secure web-based database and software application that supports various processes of public sector information systems and that stores systematic and reliable metadata about public databases. According to statistics [12], at the end of 2009, RIHA contained the data of 450 databases and IS-s. Another important tool currently under development is SEHKE [15] that is a webbased collaborative environment for creating, managing and annotating semantic resources of state IS-s and registries. The semantic interoperability architecture was set up and supported by related legislation in Estonia in 2008 demanding from holders of state IS-s creation of corresponding domain ontologies and semantic annotations of web-services. However, in reality this is not happening so easily. There is lack of knowledge about semantic technologies in general and about ontology engineering in particular. Within the semantic interoperability framework a large number of training courses have been provided to stakeholders (i.e. approximately to 200 participants), mainly to domain experts, who are responsible for ontology creation in their respective domains. A feedback from the very first training courses already has shown a need for a methodology and guidelines for ontology development that are closer to domain experts than general existing methodologies. In public administration agencies, it was impossible to have a large number of ontology engineers or knowledge engineers to help to convert domain knowledge to a formal ontology. The practical methodology for development of e-government domain ontologies presented in this paper is motivated by the situation described above.

4

A Practical Ontology Development Methodology

4.1

Goal, Object and Approach of the Methodology

A practical ontology development methodology presented in this paper aims at providing a systematic approach to development of ontologies capturing domain knowledge and semantics of databases fields related to different state IS-s. Most important goal of ontology development is their reuse across many applications and for building other ontologies.

6 7

SAWSDL, http://www.w3.org/TR/sawsdl/ XMI, http://www.omg.org/spec/XMI/

A Practical Methodology for Development of a Network

5

From the methodological point of view, development of ontologies is a complex task and a lot of ontology building experiences are needed in order to produce well established ontologies. Therefore, one of the goals of current methodology is to make ontology development process easier for domain experts by providing processes and guidelines that they could use. The provided methodology is primarily intended to be used for development of ontologies represented in OWL enabling semantic annotation of database objects and e-government services (described in WSDL) of state IS-s. In reality, domain ontologies of state IS-s form a network of ontologies, in what different domain ontologies are related by linking, modularity, versioning or by other relationships. Such ontology network creates certain requirements to ontology development methodology. Ontology maintenance becomes much more complex in the case of ontology network than in the case of isolated ontologies. Concerning methods for building domain ontologies, already accepted ontology development methodologies like METHONTOLOGY [4] and NeOn [5] are taken as a basis for creation of a new methodology that better meets specific needs of domain experts. 4.2

Ontology Development Process

The ontology development process refers to activities that are performed when building ontologies as defined in METHONTOLOGY methodology for ontology construction [4]. Ontology development process proposed in this paper is based on this definition. It takes into account as an input different re-useable knowledge resources that are available in governmental agencies, who are responsible of their IS as well as resources managed by developers of those IS. For example, re-useable knowledge resources could be regulations of creation of a state IS, conceptual schema of a state IS database, data structures in the form of XMI file, descriptions of data services etc. For creation of a domain ontology for each of the state IS domains an iterative life cycle model was chosen. According to this model ontology development starts from partial satisfaction of requirements. During each of the iterations ontologies will be improved until all the requirements will be met. This life cycle model was successfully used in development of many domain ontologies of Estonian state IS including the most important and complex ones as ontology of central commercial register8, address objects, mineral deposits, Estonian topographic database etc. In development of a network of domain ontologies of state IS a method of evolutionary prototyping is proposed. According to this a partial result (a partial network of state IS domain ontologies) that meets known requirements is developed at first. This prototype of ontology network is evaluated and assessed by using it in different applications and the requirements are refined based on such evaluations.

8

https://riha.eesti.ee/riha/onto/ettevotlus.ariregister/2010/r2

6

4.3

H.-M. Haav

(Re)usable Knowledge Resources

Reusable knowledge resources are formed from general and specific domain knowledge, non-ontological and ontological resources. Reuse of these resources speeds up ontology development process. Non-ontological resources are conceptual schemas of databases, glossaries, vocabularies, thesauri, regulatory documents of a state IS, descriptions of database objects, databases, service descriptions, etc. Ontological resources are existing ontologies in RDF or OWL format available at ontology libraries or repositories. For example, domain ontologies of state IS collected to RIHA repository could be reused for construction of a new ontology in ontology network. 4.4

Sequence of Activities

Sequence of ontology development activities is depicted in Figure 1. Management and support activities: knowledge acquisition, evaluation, improvement, management, feasibility study, documentation, etc,

Ontology development Conceptualization

Specification

Collection of central concepts

Concept generalization and specification

Addition of relationships

Implementation: addition of attributes

Ontology in OWL Implementation: addition of concepts and relationships

Resource re-engineering

Ontology reuse Resource reuse

Non-ontological resources

Ontological resources

Fig. 1. Development process of domain ontologies for state IS: sequence of activities

Ontology management and support activities are performed in parallel with other ontology development activities during the whole ontology development process.

A Practical Methodology for Development of a Network

7

Figure 1 shows also relationships of non-ontological and ontological resources with development process through corresponding activities. Main activities of ontology development process are given in the Figure 1 as the following sequence: specification, conceptualization and implementation. According to iterative life cycle model early implementation of domain concepts and relationships is foreseen. At the first iteration, basic (central) domain concepts and relationships are arranged to ontological structure and implemented. After that, during each of the iterations, new concepts and relationships are added and implemented. Attributes of individuals of concepts are added in the final stage. After that ontology is evaluated according to the requirements e.g. are all input/output parameters of web services covered by corresponding components of respective domain ontology?. If not, then missing components need to be added. In the following sections the ontology development process and corresponding activities are provided in detail. Management and Support Activities Management and support activities are similar to those defined in METHONTOLOGY methodology [4]. Ontology development starts with activity of knowledge acquisition that will continue during the whole development process. Most of the knowledge will be acquired in the beginning of the process; needed amount of new knowledge is declining according to the development process proceeds. Reuse of Resources We consider ontological and non-ontological resources as stated above. In the development process of state IS domain ontologies both types of resources could be used. Reuse of non-ontological resources is most important at the stages of ontology specification and conceptualization. Reuse of ontological resources is not possible only during specification and conceptualization but also during the implementation activity. For example, ontology import is a part of implementation of ontology. Resource reuse activities are dependent on its type as follows. In case of non-ontological resources first resource reuse is performed and then resource reengineering activity is carried out. For example, a domain model of an IS can be easily used for acquiring possible classes to be included to corresponding domain ontology. Descriptions of data objects from the corresponding XMI file can be reused for defining data type properties for individuals of classes. In the process of non-ontological resource (e.g. database schema) reuse the resource is analyzed with respect to its usability for ontology creation. If reusable resources are found, then reengineering of these resources is carried out. This is the process of transforming an existing non-ontological resource (e.g. data base schema, database etc.) into ontology that can be reused in the process of ontology engineering of current domain ontology. In case of ontological resources, reengineering of some existing ontology is performed. An example is geometry ontology that can be reused in a number of ontologies related to the domain of land IS-s.

8

H.-M. Haav

Process of ontology reuse can be divided into the following activities: 1. Retrieval of possible candidate ontologies from public ontology repositories like Swoogle9 and WATSON10. Ontologies developed for Estonian state IS can be retrieved from RIHA11. 2. Evaluation of retrieved candidate ontologies from the point of view of requirements (content, language, formal representation, documentation, quality, etc). 3. Choice of suitable ontology 4. Deciding a way how ontology is to be reused. Ontology can be reused as it is. Ontology might need reengineering before it can be reused. Ontology reengineering is the process during what its conceptual model is transformed into a new conceptual model that meets requirements of a new ontology and which is then implemented. Ontology could also be merged to a new ontology. Specification In parallel with knowledge acquisition a set of requirements that ontology should meet are to be defined. In the process of specification activity, goal and scope of ontology, resources, and use cases are specified. In addition, a list and descriptions of central domain concepts are provided. For the development process specification helps to stay within the frame of the scope of the problem. For third parties, ontology specification serves as an initial overview of ontology. Conceptualization and an Early Implementation Ontology specification activity is followed by ontology conceptualization. During the conceptualization, a conceptual model of ontology that represents domain concepts and relationships between them is created. The conceptual model of ontology should take into account that ontology will be later implemented in OWL language. Ontology conceptualization is carried out according to acquired knowledge of a domain. Ontology conceptualization is followed by ontology implementation activity resulting in ontology description of OWL language. In provided methodology ontology conceptualization and implementation are considered together because iterative life cycle model of ontology development is applied. Therefore, we call implementation stage here as an early implementation. Different strategies could be used for conceptualization as follows: • Top-down strategy starts from most general domain concepts and step-by-step moves to specific concepts. • Middle out strategy proposes at first to concentrate to a set of basic (central) concepts of a domain and then to move to the direction of their generalization and specialization. • Buttom-up strategy tries to generalize meaning of data fields related to an application/service. 9

Swoogle; http://swoogle.umbc.edu/ Watson; http://watson.kmi.open.ac.uk/WatsonWUI/ 11 RIHA; http://riha.eesti.ee 10

A Practical Methodology for Development of a Network

9

Taking into account that domain ontologies for state IS will be created by domain experts on the basis of their data collections and corresponding IS the middle out conceptualization strategy is the most appropriate. Central concepts can be identified easily by domain experts as these are usually denoted by central terms used to talk about data collection of the corresponding IS. These terms are also used in legislation documents related to IS. Using top-down approach (from legislation terminology to data fields) creates too many concept hierarchy levels to ontology. Bottom-up method makes it hard to generalize from the level of data fields to more general concepts and therefore it takes more time to distinguish between concepts and attributes. This in turn means that development of ontology that satisfies requirements takes more time. In proposed methodology ontology conceptualization and implementation are divided into sub-activities as follows: 1. Collection of concepts. Central concepts of a domain are identified (for example, 7-10 concepts). Purpose and scope of IS, most often used database fields and service parameters are helpful indicators for identifying central domain concepts of ontology. 2. Concept generalization and specification of one level. Not all concepts of ontology should correspond to input/output parameters of services, there could be concepts that make ontology better understandable for third parties. However, it is recommendable to keep ontology as minimal as possible with respect to its scope. 3. Identification and addition of main relationships. After central concepts and their generalizations and specifications are identified, then the main relationships between these concepts are to be added to ontology. 4. Early implementation. Result of ontology implementation is its description in OWL. Early implementation enables to detect logical errors of ontology description as well as get a view of its complexity and see how well it meets the requirements. As a rule, ontology creation tools like Protégé12 are used. These tools include built-in DL reasoners (for example, Pellet13 , Fact++14 etc) in order to automatically find inconsistencies in ontology description presented in OWL. 5. Iterative development. As the main goal of ontology its to cover all concepts related to input/output parameters of services, then during consecutive iterations all service parameters will be taken into account and corresponding to them concepts are identified and step-by-step added to ontology. In case of iterative development a set of new concepts to be added to ontology is decreasing making possible to limit development process to implementation activity in the final iteration steps. 6. Addition of datatype properties. As a rule, datatype properties correspond to data objects (database fields like last name, first name, SSN, etc) of a given state IS and are added to the ontology and implemented during final iteration of its development.

12

Protégé; http://protege.stanford.edu/ Pellet; http://www.clarkparsia.com/pellet 14 Fact++; http://owl.man.ac.uk/factplusplus 13

10

H.-M. Haav

7. Annotation of ontology components. After logical consistency of ontology is proved ontology components are to be annotated (e.g. descriptions of natural language(s) are added to components). It is user-friendly to provide a documentation of ontology in addition to its description in the form of machinereadable OWL file. This makes it easier to reuse, maintain and further development of ontology. In order to create conceptual model of ontology it is recommended to use some method of knowledge organization and structuring. Glossary of terms and concepts, concept taxonomy, list of relationships between concepts and attributes of individuals could serve for this purpose. Corresponding activity could be called as documentation of conceptualization of ontology.

5

Evaluation of the Methodology

The provided methodology has been iteratively developed and evaluated during the last 2 years within the framework of achieving semantic interoperability of state IS in Estonia. The methodology was widely used in numerous training courses on ontology engineering provided for domain experts of state IS-s. The feedback from participants of courses was used to improve the methodology in order to meet needs and take into account background knowledge of domain experts. By now, the methodology is accepted as the ontology development methodology for creation of domain ontologies of state IS in Estonia. During the year 2010, 21 domain ontologies of state IS were developed using this methodology. The complexity of ontologies required for achieving semantic interoperability of state IS as already mentioned is not very high. Domain ontologies of state IS are intended to be used for semantic annotation of web services and as a rule do not contain descriptions of defined classes. The following table 1 represents some quantitative characteristics of a few typical state IS domain ontologies. Table 1. Ontology metrics of some state IS ontologies Name of ontology

Address objects Central commercial register Mineral deposits Estonian topographic domain Estonian geodetic domain

Data-type properties

Subclass axioms

13 39

48 147

17 100

ALCQ(D) ALQOIQ(D)

25 159

19 9

133 39

12 158

ALCQ(D) ALCRQ(D)

16

23

58

0

ALU(D)

Count of classes 17 101

Object properties

DL expressivity

A Practical Methodology for Development of a Network

11

It was intention from the very beginning of the ontology development for state IS to keep ontologies as simple as possible in order to make it possible to domain experts to create ontologies of their respective fields. Some of ontologies are not entirely finalized as more datatype properties are to be added in order to meet requirements of semantic annotations of web services. According to the law, domain ontologies are to be collected to RIHA repository. However, not all of them are there. In 2010, only 7 ontologies were published. A quality assurance procedure of ontologies to be stored in RIHA has put in place and before ontology does not meet respective requirements, it is not published in RIHA. The methodology proposed in this paper aimed at development of ontology network of state IS domain ontologies. With this respect RIHA serves as ontology repository for ontology reuse and re-engineering. The SEHKE tool [15] is intended to support collaborative ontology development and managing network of ontologies. State IS in Estonia has according to RIHA over 20 000 data objects. These should be annotated using corresponding ontology components (i.e. datatype properties). Consequently, the ontology network should contain approximately the same number of datatype properties in addition to concepts and object properties. By now, we have about 500 datatype properties defined in our current version of the ontology network. According to this, we are in the beginning of development of ontology network of state IS in Estonia. These data objects are used by about 1500 X-road services (machine-to-machine data services) which input/output parameters should be semantically annotated. This process has just started.

6

Conclusion

Demand for state IS domain ontologies to be used for solving semantic interoperability problems of e-government services requires easier process of creating ontologies than general technically complex ontology development methodologies provide. In this paper we presented a novel ontology development methodology for domain experts who being actual domain knowledge holders easily identify central domain concepts and using early implementation and iterative development activity could build ontologies without need for ontology engineer. The approach is evaluated by development of network of e-government domain ontologies of state IS in Estonia. Lessons learned from the use of the methodology show that implementation activity is sometimes still too complex for domain experts even if they use easy-to-use ontology editors. Therefore, the future work is planned to make this ontology development methodology even easier by using simple intermediate representations of domain conceptualization that can be automatically translated to OWL ontology. In this case, implementation activity of provided methodology will be automatic or will be at least semi-automatic. Acknowledgments. This research was partially supported by the target-financed theme No. 0322709s06 of the Estonian Ministry of Education and Research as well as

12

H.-M. Haav

by the ERDF funded Estonian Centre of Excellence in Computer Science, EXCS. I am particularly grateful to Leive Turi for creating the first ontology using this methodology and collaborating to the development of the methodology.

References 1. Barzdins, J., Barzdins, G., Balodis, R., Cerans, K., Kalnins, A., Opmanis, M., Podnieks, K.: Towards Semantic Latvia. In: Proceedings of Communications of 7th International Baltic Conference on Databases and Information Systems, Vilnius, Lithuania, pp. 203–218 (2006), http://melnais.mii.lu.lv/audris/CReady_Barzdins_TowardsSeman ticLatvia.pdf 2. Bednár, P., Furdík, K., Paralič, M., Sabol, T., Skokan, M.: Semantic Integration of Government Services - the Access-eGov Approach. In: Cunningham, P., Cunningham, M. (eds.) Collaboration and the Knowledge Economy: Issues, Applications, Case Studies. IOS Press, Amsterdam (2008) 3. Georgiadis, P., Lepouras, G., Vassilakis, C., Boukis, G., Tambouris, T., Gorilas, S., Davenport, E., Macintosh, A., Fraser, J., Lochhead, D.: SmartGov: A Knowledge-based Platform for Transactional Electronic Services. Springer, Heidelberg (2002) 4. Gómez-Pérez, A., Fernández-López, M., Corcho, O.: Ontological Engineering with examples from the areas of Knowledge Management, e-Commerce and the Semantic Web. Springer, Heidelberg (2004) 5. Haase, P., Rudolph, S., et al.: NeOn Deliverable D1.1.1 Networked Ontology Model (November 2006), http://www.neon-project.org/ 6. Haav, H.-M., Kalja, A., Küngas, P., Luts, M.: Ensuring Large-Scale Semantic Interoperability: The Estonian Public Sector’s Case Study. In: Haav, H.-M., Kalja, A. (eds.) Databases and Information Systems V, pp. 117–129. IOS Press (2009) 7. Hyvönen, E., Viljanen, K., Tuominen, J., Seppälä, K.: Building a National Semantic Web Ontology and Ontology Service Infrastructure –The FinnONTO Approach. In: Bechhofer, S., Hauswirth, M., Hoffmann, J., Koubarakis, M. (eds.) ESWC 2008. LNCS, vol. 5021, pp. 95–109. Springer, Heidelberg (2008), http://www.seco.tkk/publications/2008/hyvonen-et-albuilding-2008.pdf 8. IEEE Standard for Developing Software Life Cycle Processes. IEEE Std 1074-1997 9. Kalja, A.: New version of the x-road. In: Information Society Yearbook 2009, Ministry of Economic Affairs and Communications of Estonia, Department of State Information Systems (RISO), Tallinn (2010), http://www.riso.ee/en/pub/2009it/#p=21-4 10. Lavbiĉ, D., Krisper, M.: Rapid Ontology Development. In: Welzer Druzovec, T., et al. (eds.) Proceeding of the 2010 Conference on Information Modelling and Knowledge Bases XXI, pp. 283–290. IOS Press (2010) 11. Loutas, N., Tarabanis, V.P.: Providing public services to citizens at the national and paneuropean level using semantic web technologies. In: Proceedings of the Sixth Eastern European eGov Days, Czech Republic (2008) 12. Parmakson, P., Vegmann, E.: The administration system of the state information system (RIHA). In: Information Society Yearbook 2009, Ministry of Economic Affairs and Communications of Estonia, Department of State Information Systems (RISO), Tallinn (2010), http://www.riso.ee/en/pub/2009it/#p=2-1-6

A Practical Methodology for Development of a Network

13

13. Staab, S., Schnurr, H.P., Studer, R., Sure, Y.: Knowledge Processes and Ontologies. IEEE Intelligent Systems 16(1), 26–34 (2001) 14. Stojanovic, L., Kavadias, G., Apostolou, D., Probst, F., Hinkelmann, K.: Ontology-enabled e-Gov Service Configuration, Deliverable Ontogov Project - IST Project 507237, http://www.ontogov.com 15. Tomingas, K., Luts, M.: Semantic Interoperability Framework for Estonian Public Sector’s E-Services Integration. In: d’Aquin, M., García Castro, A., Lange, C., Viljanen, K. (eds.) ESWC 2010/ORES 2010 Proceedings of the First Workshop on Ontology Repositories and Editors for the Semantic Web (ORES 2010) 7th Extended Semantic Web Conference Hersonissos, Crete, Greece (May/June 2010) 16. Uschold, M., Grueninger, M.: Ontologies: principles, methods and applications. Knowledge Sharing and Review 11 (1996) 17. Zhou, L., Booker, Q., Zhang, D.: Toward Rapid Ontology Development for Underdeveloped Domains. In: Proceedings of the 35th Annual Hawaii International Conference on System Sciences (HICSS 2002), vol. 4, pp. 106–115 (2002)

Channel Choice in the G2B Context: Towards a Research Agenda Yvon van den Boer, Willem Pieterson, and Rex Arendsen University of Twente, Center for e-Government Studies P.O. Box 217, 7500 AE Enschede, The Netherlands {y.vandenboer,w.j.pieterson,r.arendsen}@utwente.nl

Abstract. Knowledge about e-governmental service delivery towards businesses in general and the use and choice of service channels in particular is lacking. Current insights are mostly based on research in the context of citizens. Important differences in both domains can be found in the more networked character of the business context. These differences show that research towards the business context is needed. In order to fill the gap we provide an overview of existing knowledge about channel choice in G2B context. To conclude, we present a research agenda for future research on this topic. Keywords: channel choice, G2B interaction, service delivery, network perspective.

1

Introduction

Governmental organizations deliver services to citizens and businesses. Due to higher administrative burdens [1] and the complexity of their relationship with governments, businesses have more contacts with governmental agencies than citizens [1]. Given the importance of e-Services [3] for businesses, the European Commission aims that by 2015, 80% of enterprises will have used e-government. In addition, businesses are expecting high quality in the governmental public service delivery [4]. Therefore, one would expect a large amount of attention in e-government research for the service delivery towards businesses in general, and the use and choice of service channels in particular. However, most of the research in channel choice focuses on citizens [5-7]. Given the complexity of the Government to Business (G2B) relationship, it is questionable whether the findings of Government to Citizens (G2C) studies are fully applicable to the G2B context. This observation calls for insights on channel choice in the G2B context. However, almost no studies exist that examine entrepreneurial channel choice behavior. In order to fill this hole and reach the targets of the European Commission [3], it is important to gain more knowledge about the channel choice behavior of businesses. The goal of this paper is to review the existing literature on channel choice in the G2B context to provide an overview of the current knowledge and the gaps in the scientific literature. Given the fact that the body of literature in the governmentcitizens context is more developed, we assess the differences between the domains to establish a research agenda for channel choice research in the G2B domain. This leads to the following research questions: T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 14–27, 2011. © IFIP International Federation for Information Processing 2011

Channel Choice in the G2B Context: Towards a Research Agenda

RQ1a: RQ1b: RQ2: RQ3: RQ4:

15

What is the current state of the art in theories of channel choice in the G2B context? What is the current state of the art in research of channel choice in the G2B context? How does the existing knowledge about channel choice in the G2B context differ from the knowledge in the G2C context? How do the differences lead to different insights regarding channel strategies? What are the main topics for future research about channel choice in the governmental public service delivery to businesses?

Based on our analysis and the answers to the research questions, we formulate a number of statements that form the basis of the proposed research agenda. The first section of this paper gives an overview of the backgrounds of channel choice and provides an answer to RQ1a. The second section of this paper discusses the existing studies on channel choice of businesses and answers RQ1b. After that, at the end of the first two sections, an answer will be provided on the second and third research question. The final section discusses possibly interesting topics for upcoming research and will also present a research agenda, which also delivers an answer on RQ4.

2

Backgrounds of Channel Choice

In this section we discuss theories of channel choice, focused on both citizens and businesses, in order to answer the first research question (RQ1a) of this paper. 2.1

Channel Choice in the G2B Context

Before we look at channel choice behavior of businesses, we have to provide some definitions of channel choice and apply these to the context of businesses. A channel can be defined as “a customer contact point or a medium through which the firm and the customer interact” [8, p. 96]. In this research, we apply the same definition to G2B interactions. So, the firm in the definition is the governmental agency and the customer is a business. Channel choice is defined as “the choice of the customer to use a specific channel in a specific situation” [9, p. 9]. Again, the customer can be seen as a business in this context. We see channel choice as a bureaucratic confrontation between businesses and governmental agencies [10]. Directing channel choice is defined as “the use of instruments to direct the behavior of citizens in the desired course” [11, p. 10]. In accordance with the other definitions, we can see citizens as businesses in the context of G2B interaction. Subsequently it is important to define a business. Jaffee [12] describes a business as a network of individuals. This network of individuals can be partially influenced and depends on the environment. In addition, a business can be an individual or a group of individuals showing particular behavior [13]. Besides, there is the law of individual differences [14]: 1) Individuals within a business differ in background, experiences, perceptions and expectancies, and 2) individuals react differently to situations. Last of all, we define (communication) networks as: “the patterns of contact that are created by the flow of messages among communicators through time and space” [15, p.3].

16

Y. van den Boer, W. Pieterson, and R. Arendsen

There are multiple ways to interact with the government. Which channel is chosen depends on the type of service [10]. Table 1 shows that there are different types of channels, and that each type of channel has its own features. Table 1. Types of channels [10]

Channel type Personal Telephone Paper Electronic

Kind of interaction Face-to-face Telephonic Print media Textual Audio-visual

Example(s) Front desk, an intermediary Mail, fax Websites, e-mail Video chat

We cannot automatically apply these types of channels into the context of businesses. Hence, we have to make two important remarks about the abovementioned types of channels. The first remark is about the role of the intermediary. Changes in the society itself and changes in technology were leading towards the development of the network society [16, 17]. According to these changes G2B-relationships were getting increasingly horizontal and were no longer vertical, and information exchange is now fundamental to G2B relationships instead of power [18]. These developments changed the role of the intermediaries in the G2B context. Motivated by the use of Internet as a service provision channel new forms of intermediation were necessary [19]. Pieterson [10] presented an intermediary as a personal type of channel in G2C interaction. Others argued, that from a governmental point of view, the intermediary is not a channel but a source of information [1]. An intermediary is more like an external source of information, which can get reached through certain channels. Complexity in the issuing of laws and rules, and sometimes also complexity in technology are supporting arguments for a relationship between a business and an intermediary [12]. Other reasons, especially for small and mediumsized businesses, are an increasing focus on primary products and processes and gaining efficiency benefits [20]. Besides, Pfeffer and Salancik [21] elucidated in their resource dependence perspective that organizations are embedded in an environment with other organizations and are to some extent dependent on resources of others in that environment [21]. They are not completely self-contained. Businesses therefore specialize on their core competences and need others to fulfil secondary tasks. Those others can be seen as the external network of a business. For example, a contractor is specialized in building roads and as any other organization is obliged to fill in the taxes. The primary task of the contractor is building roads and they, usually, have little knowledge how to properly submit their taxes. Therefore, they need another organization that is specialized on accountancy. This firm of accountants is the intermediary party in G2B interaction. From a governmental point of view, the reasons to bring in intermediary organizations are reducing the amount of contact and data collecting points [22]. Intermediaries fulfil different roles in G2B interaction and can be seen as important partners in optimizing (e)governmental services towards businesses [20]. Besides these intermediary organizations (e.g. formal intermediaries), social intermediaries (e.g. friends and family) already fulfil an important role in

Channel Choice in the G2B Context: Towards a Research Agenda

17

optimizing e-government services towards citizens [23]. However, it is unknown if these social intermediaries also play an important role in the context of businesses. Although, Jansen et al. [1] distinguished different sources and they found that personal network (e.g. external advisors, friends, family, colleagues) can be important for businesses. So, an intermediary is a source of information in the G2B context. This is an important difference in comparison with the context of citizens and adds a networked character to the G2B context. The second remark is about the rise of social media as the newest form of communication space [24]. Social media can be seen as a potential new channel for egovernmental service [25] and the public sector is slowly adopting this new technology from the private sector [28]. Democratic representatives already use social media to interact with citizens, for example during campaigns [25, 26]. Social media have the potential to leverage a shift towards a more open and collaborative culture [25], as well as filling the gap between politics and citizens [26], increasing accountability [25] and increasing political and societal engagement [25]. Finally, new horizontal communication networks arise from social media and have the typical traits of the network society with regard to technical and organizational structure [24]. Hence, the number of channels and sources used in G2B interactions is higher than taken into account in current G2C studies. Besides, the current state of the art in G2C research does not reflect the actuality of businesses’ channel choices. Further, G2B interactions have a far more networked character than G2C interactions, due to the role in all kinds of intermediaries and this networked character impacts their channel behavior. 2.2

Theories of Channel Choice

There are different perspectives on channel choice. The first is a rational perspective on channel choice [9]. This perspective is based on the Media Richness Theory [29] and the Social Presence Theory [30]. It assumes that people think in a rational way when they choose a channel. A few years later, the Social Influence Model [31] and the Channel Expansion Theory [32] add a social and subjective side to the model, which implicates that there also are variables that influence channel choice unconsciously. These perspectives have often been criticized because of the lack of empirical support [11]. A recent development in channel choice research, made by Pieterson [10], has resulted in a new theory of channel choice. This theory focuses on citizens in their interaction with the government. Pieterson [10] found that the process of channel choice is not as straightforward as researchers often assumed. When citizens need to choose a channel, they primarily rely on their habits and experiences. In the second instance, there is a rational elaboration. So, citizens use different strategies to make a channel choice when they want to contact the government [10]. Which strategy is chosen depends on personal characteristics (e.g. age or education) as well as task (e.g. complexity of the goal) and channel characteristics (e.g. speed of a channel and its ease of use) [9]. In addition, situational (e.g. a persons need for closure, time and distance) [37] and emotional constraints influence the choice of a strategy. Before we continue we have to make some remarks here. The abovementioned theory of channel choice [10] focuses on citizens. It is unknown if this theory can

18

Y. van den Boer, W. Pieterson, and R. Arendsen

automatically be placed into the context of businesses. So first of all, research is needed to verify this theory on channel choice of businesses. Secondly, besides the influencing factors on channel choice Pieterson [10] determined, others argued that there are more influencing factors on channel choice. We discuss these factors and point out the most important differences in influencing factors on channel choice of businesses compared to citizens. First of all, the channel choice of businesses depends on the type of business and thereby on the type of service [9]. It is therefore not unimaginable that characteristics of the organization, such as size, industry and age of the organization have influence on their channel choice behavior. So, besides to the aforementioned influence of personal characteristics [10], the possible influence of organizational characteristics needs to be examined in research on channel choice of businesses in the G2B context. Secondly, social factors also play an important role in channel choice behavior; because a sympathizer can influence someone else to use a certain channel [33, in 8]. From a social psychology perspective, researchers [34] presume that social norms in a group have influence on the behavior of individuals. This aspect of social influence can bring a new valuable perspective. Keeping in mind that channel choice behavior is an individual action [10], we can imagine that in the business context employees influencing each other’s channel choice behavior. In addition, [35, in 12] pointed out some important aspects of organizational reality: 1) structures, like social structural elements (norms, hierarchy etc.) 2) processes that are goal directed and 3) outcomes, the consequences of both structures and processes on employees, the organization itself and the whole society. The first aspect is important for channel choice behavior because the social structural elements have influence on behavior of the individual within the organization and it shows the difference between a citizen and an employee [36]. So in this case, the group can be seen as a network of individuals that together form a business. We call this the internal network of a business. Here we can identify the distinction between the individual as a citizen and the individual as part of the internal network of a business. Compared to channel choice of citizens, individuals get influenced in their channel choice by their colleagues within the internal network, which is the business. This is the third statement we make, which adds a network perspective to channel choice of businesses and shows the importance of gaining more specific knowledge about channel choice of businesses. In order to answer the first research question (RQ1a) concerning the current state of the art of theories in channel choice of businesses, we draw these conclusions: First of all, there is little knowledge of theories in channel choice behavior of businesses in a governmental public service delivery context. Moreover, we only found a channel choice theory of Pieterson [10] concerning the channel choice behavior of citizens in a governmental service delivery context. Nevertheless, we cannot apply this theory straight into the business context, because the influence of the factors on channel choice is unknown in a business context. All in all, we found three important differences between the context of citizens and the context of businesses, which lead towards the following statements: St1 There are more channels and sources to take into account in G2B interaction compared to the current knowledge of channel choice in G2C interaction.

Channel Choice in the G2B Context: Towards a Research Agenda

19

St2 The context of G2B interaction seems to have a more complex and networked character than the context of G2C interaction, because of the presence of intermediaries. St3 The channel choice of an employee is influenced by the social norms within the internal network of a business, which gives the channel choice of businesses in a governmental service delivery context an even more networked character. These differences lead towards the overall statement that the theory about channel choice of citizens cannot directly be applied to the business context. So, further research is needed to identify all the factors of influence regarding channel choice behavior of businesses. In the next section we will find out if we can gain more information about the factors influencing channel choice of businesses.

3

Research of Channel Choice in the G2B Context

In this section we review the few studies that have been conducted on businesses’ channel choice (Table 2). In this section we discuss these findings to provide an answer on RQ1b, RQ2 and RQ3. Table 2. Overview of existing studies of channel choice behavior of businesses

Subject Channel preference and/or use

Motivations to use a channel

Perceptions of channels Functions of channels Source choice

Authors De Vos [39] Van den Boom [40] Jansen & Pieterson [36] Van de Wijngaert et al. [4] De Vos [39] Van den Boom [40] Van de Wijngaert et al. [4] De Vos [39] Van de Wijngaert et al. [4] Jansen et al. [1]

N 115 601 323 323 115 601 323 115 323 323

In this analysis, we focus on five key aspects of businesses’ channel behavior; 1) general use and channel preference, 2) the reasons businesses have to choose a certain channel, 3) their perceptions of service channels, 4) the functions of channels and finally, 5) the selection of a certain information source. First and foremost, researchers found that channel preference does not automatically lead to the use of that preferred channel [39, 36, 4]. So, intentions are not always determined for behavior [36]. More specifically, the most preferred channel of businesses in a governmental context is the telephone, closely followed by the Internet [39,36]. However, Van den Boom [40] found that email is preferred above the website. Here, the telephone is one of the least used channels. On the other hand, the channel most used by businesses is the telephone with the Internet in second place [39, 36, 4]. So, there seems to be a switch between the telephone and Internet regarding preference and use of channels (Table 3).

20

Y. van den Boer, W. Pieterson, and R. Arendsen Table 3. Channel preference versus Channel use

Authors De Vos [39] Jansen & Pieterson [36]

Most preferred channel 1. Telephone 2. Internet 1. Telephone 2. Internet

% 42 41 65 21

Most used channel 1. Internet 2. Telephone 1. Internet 2. Telephone

% 49 39 53 30

The second point concerns the motivations to use a channel. The most important motivations in general are ease and speed [39]. For the phone, the design plays a role. The motivations to choose the desk are design and certainty. In addition to this, others found that other motivations to choose a certain channel are goal and situation [40, 4]. We have to remark that these findings are based on questions about intentional behavior and are not based on actual behavior. As we saw before in this paper, intention and behavior are not always the same [13]. There also is some knowledge about the perceptions of channels, from a business point of view. Businesses perceive the telephone as an appropriate channel for clarifying situations, using different languages, having a quick answer, reaching their goal and to obtain the best service and experience [39]. Businesses also perceive the Internet as a cheap channel, appropriate for the searching of information. However, when we compare the perceptions of channels from businesses [39] and citizens [9], we see a difference in perceptions of channels (figure 2.). Channel perceptions vary largely between the different groups, leading to different possibilities for service delivery.

Fig. 2. Channel perceptions: citizens [9] versus businesses [39]

Channel Choice in the G2B Context: Towards a Research Agenda

21

An explanation for the different perceptions can be found in the difference of the tasks in both contexts because of different service delivery [36]. Another important explanation can be derived from the more networked character at business level of G2B interaction, as we discussed in the theory section, compared to G2C interaction. More specifically, the presence of the internal network of businesses can lead towards different perceptions of channels by an individual as part of a business instead of being a citizen. For example, the individual as a citizen would go to the front desk for an answer on a specific question. In contrast, the individual as part of a business would use the telephone to get an answer on that question because all colleagues are doing the same and it is not common to go to the desk. In this case, the telephone is perceived as more useful in the business context instead of the front desk in the citizens’ context. So, this leads to the statement that the channel perception of an employee is influenced by the social norm within the internal network (e.g. colleagues) of the business. Regarding the fourth point, there is little knowledge about the functions businesses attribute to the channels Internet, telephone and desk. Businesses use the Internet as a channel when they have a general and simple question [4]. When questions get more specific and are more important for the businesses, they mostly use the telephone or the desk as a channel to interact with the government. Besides, research shows that citizens sometimes use different channels to reach a specific goal [38]. For example, a citizen first looks for more information on a website, and afterwards he or she verifies the information at a service desk with faceto-face communication. Furthermore, others found that businesses also use different channels when they contact the government [4]. Finally, Jansen et al. [1] researched the source choice of businesses in a public service context. They distinguished between sources in ones personal network (e.g. external advisors, family and friends, internal and external colleagues), expert organizations (e.g. chamber of commerce, professional organizations), government organizations (e.g. national government and municipalities), portals and search engines. They found that compared to smaller businesses, larger businesses make less use of expert organizations and more use of their personal network. Finally, they also found that importance of the issue and the needed specificity of the information are important factors for source choice. To conclude this section, we formulate an answer on the first research question (RQ1b): What is the current state of the art in research of channel choice in the G2B context? Presently, there are a few studies that researched the channel choice of businesses in the governmental public service delivery context. There is very little knowledge about the factors that influence channel choice of businesses. Moreover, most studies examined the measurable outcomes of channel choice, like channel preference and use, and provided some possible explanations of their findings. So, the current studies are just an exploration of channel choice of businesses and thorough research is needed. However, one remarkable finding is worth to mention here: the intention to use a certain channel, does not automatically lead to the use of that preferred channel [39, 36, 4].

22

Y. van den Boer, W. Pieterson, and R. Arendsen

The second research question (RQ2) was: How does the existing knowledge about channel choice in the G2B context differ from the knowledge in the G2C context? We found two important differences between channel choice of businesses and citizens that leads to the following statements: St4 Businesses and citizens have different perceptions of channels. An explanation can be derived from the theory discussed in the first section. Compared to citizens, individuals as employees are influenced in their channel choice by an internal network (e.g. colleagues) of a business. There is a difference in context. St5 Source of information seems to be a more important factor in the understanding of channel choice of businesses compared to citizens. The reason for this can be found in the more networked character of the business context compared to the context of citizens. In this case, we mean the external network of the business in their interaction with the government during service delivery. With this in mind, we can formulate an answer on the third research question (RQ3): How do the differences lead to different insights regarding channel strategies? So far, our study illustrates that the context of channel choice in governmental public service delivery to businesses is different from that to citizens, as the various statements make clear. So, the development of a multichannel strategy in channel choice for G2B interaction seems to be highly desirable. In this way, businesses will choose the right channel for the right service. However, the current theory and research about channel choice of businesses is very scarce. In the next section of this paper we will provide a research agenda on this topic in order to extend the knowledge of channel choice of businesses in the near future.

4

Conclusion: A Research Agenda

The goal of this paper is to review the existing literature on channel choice in the G2B context in order to provide an overview of the current knowledge and the gaps in the scientific literature. Our study shows that theories in the G2C context cannot directly be applied to the G2B context, because of crucial differences between both domains. Besides, the few studies conducted on G2B interaction are merely an exploration and thorough research is needed. In this section we give an overview of the statements and present research topics for future research in G2B interaction, which answers RQ4. The first topic on the research agenda focuses on the number of channels and sources used in G2B interaction. The following statement was presented: St1 There are more channels and sources to take into account in G2B interaction compared to the current knowledge of channel choice in G2C interaction. First, keeping in mind that intermediaries can be seen as sources of information instead of a channel in G2B interaction, it is desirable to gain more insight regarding different kinds of sources in order to determine their influence on channel choice of businesses. Secondly, there is little knowledge about social media as potential new

Channel Choice in the G2B Context: Towards a Research Agenda

23

channel in G2B service delivery. Hence, it is essential to extend the understanding on: 1) the potential role of social media in G2B interaction during service delivery and 2) the current role of social media in other forms of interaction with the environment of businesses. In order to estimate the potential role of social media in the G2B context it is desirable to focus on the characteristics of social media in terms of richness (e.g. task/media fit) for example [29]. Besides, perceived characteristics (e.g. relative advantage, compatibility) of social media could give more understanding about this topic [42]. The second topic on the research agenda considers the influence of the external network of businesses on their choice behavior. The following statement was formulated: St2 The context of G2B interaction seems to have a more complex and networked character than the context of G2C interaction, because of the presence of intermediaries As we discussed before, it is preferable to bring in intermediary organizations, from both business and governmental perspectives. Source choice is a key factor here. From the governmental perspective it will be desirable to identify important players in the network in order to become aware of the influence of intermediaries on the channel choice of businesses. However, there is little insight regarding the interaction between intermediaries and businesses and how it is influencing channel choice of businesses. In order to fill this research gap it is essential to gain knowledge about the external network of businesses and determine the influencing factors from the environment on the channel choice of businesses. Hence, a relevant perspective for future studies regarding channel choice behavior in the G2B context is to consider the interaction between businesses, intermediaries and governments as a social network. The (social) network analysis method can provide insight in the influence of relations between businesses and intermediaries on channel choice behavior in G2B interaction during service delivery [43]. Factors like structural equivalence, proximity, symmetry and trust are meaningful aspects in understanding and describing a network and the influences of relations on business behavior [15]. The third topic of this future research agenda considers the internal network of businesses. We presented the following statement: St3 The channel choice of an employee is influenced by the social norms within the internal network of a business, which gives the channel choice of businesses in a governmental service delivery context an even more networked character. The current literature provides little insight into the influencing factors on channel choice of employees. It is unknown to what extent business channel choice behavior is caused by individuals on the one hand, and the business itself on the other hand. For example, do employees have high autonomy to make their own choices, or are there rules for choosing a channel to contact the government? In other words, what are the influencing factors on the realization of business channel choice behavior? Relevant factors here seem to be social influence (e.g. direct statements, vicarious learning, norms for media behavior) and situational factors (e.g. individual

24

Y. van den Boer, W. Pieterson, and R. Arendsen

differences, facilitating factors, constraints) [31], in order to determine perceived influence of colleagues and supervisors on channel choice of an employee. Concerning the abovementioned topics for future research the social influence model of Fulk et al. [31] seems to be the most relevant theory for explaining channel choice of businesses. Furthermore, the theory of channel choice of citizens by Pieterson [10] could bring in some significant influencing factors regarding channel choice of businesses. However, it is not the case that existing theories provides us a complete view of the formation of today’s business’ channel choice behavior. Hence, personal characteristics (e.g. age, education) [10] and characteristics of businesses (e.g. size, industry and type of business) [2] should be taken into account when businesses choose their channel in order to contact the government. Especially the presented network perspective can be of added value in order to understand channel choice of businesses. Monge and Contractor [15] made a strong argument for multilevel theories of communication networks, which seems applicable to the G2B context, due to the possibility of linking several levels of networks. This network perspective can give more insight in the influence relationships with colleagues (internal network, organizational level) and other organizations (external network, interorganizational level) have on channel choice behavior of businesses in the context of G2B interaction. Thus, characteristics of the internal as well as the external network of businesses should be topics of interest. Regarding the fourth research question (What are the main topics for future research about channel choice in the governmental public service delivery to businesses?) we can distinguish three main research topics: 1) the number of channels and sources to take into account in G2B interaction, 2) the role of the external network, with the presence of intermediaries, on the realization of business channel choice and 3) the role of the internal network in the realization of business channel choice. In order to examine these topics of future research the first step would be to explore the determinants of business’ channel choice. The next phase would be to empirically validate these antecedents in order to conceptualize a model that explains channel choice behavior of businesses. To summarize, the networked character of the business context indicates to be the factor that makes it a complex field, but it also seems to be the key towards understanding channel choice behavior of businesses. Therefore, the network perspective is of great added value in order to understand and predict channel choice behavior of businesses. Moreover, with regard to the ambitions of the European Commission [3] it is even of greater value to extend the knowledge on channel choice behavior in order to optimize (e-)governmental service delivery to businesses.

5

Implications

Besides the presented research agenda, we can already provide some implications for governments in order to take the first step towards the understanding of business channel choice. First, the implication for governments is that the current channel strategies of citizens are not suitable to the context of businesses. Besides, it is essential to pay attention to the role of the intermediaries as a source in government-business interaction. In particular, a distinction between social and

Channel Choice in the G2B Context: Towards a Research Agenda

25

formal intermediaries possibly delivers a better understanding of different influences on choice behavior of businesses. Finally, the role of networks related to the intermediaries, are crucial in understanding business choice behavior. For these reasons, a multichannel strategy in channel choice of businesses in governmental service delivery is highly desirable.

References 1. Arendsen, R., Van Engers, T.V., TeVelde, R.: An Empirical Study on Business-toGovernment Data Exchange Strategies to Reduce the Administrative Costs for Businesses. In: Suomi, R., et al. (eds.) Project E-Society: Building Bricks, 6th IFIP Conference on eCommerce, e-Business and e-Government, vol. 226, pp. 311–323. Springer, Boston (2006) 2. Jansen, J., Van de Wijngaert, L., Pieterson, W.: Channel Choice and Source Choice of Entrepreneurs in a Public Organizational Context: The Dutch Case. In: Wimmer, M.A., Chappelet, J.-L., Janssen, M., Scholl, H.J. (eds.) EGOV 2010. LNCS, vol. 6228, pp. 144– 155. Springer, Heidelberg (2010) 3. European Commission: The European eGovernment Action Plan 2011-2015: Harnessing ICT to promote smart, sustainable & innovative Government. European Commission, Brussels (2010) 4. van de Wijngaert, L., Pieterson, W., Jansen, J., van Dijk, J.: Kanaalkeuze en bronkeuze van ondernemers in de publieke context en de rol van Antwoord© voor bedrijven daarin. Universiteit Twente, Enschede (2010) 5. Pieterson, W., Ebbers, W.: The use of service channels by citizens in the Netherlands: Implications for multi-channel management. International Review of Administrative Sciences 74, 95–110 (2008) 6. Reddick, C.G.: Citizens interactions with e-government: From the streets to servers? Government Information Quarterly 22, 38–57 (2005) 7. Thomas, J.C., Streib, G.: The new face of government: Citizens-initiated contacts in the era of e-government. Journal of Public Administration Research and Theory 13(1), 83–102 (2003) 8. Neslin, S.A., Grewal, D., Leghorn, R., Shankar, V., Teerling, M.L., Thomas, J.S., Verhoef, P.C.: Challenges and opportunities in multichannel customer management. Journal of Services Research 9(2), 95–112 (2006) 9. Pieterson, W., Teerling, M.L.: Kanaalkeuze: klanten kiezenkanalen. Novay, Enschede (2008) 10. Pieterson, W.: Channel choice: Citizens’ channel behavior and public service channel strategy. Gildeprint B.V., Enschede (2009) 11. Teerling, M.L., de Vos, H., van de Wijngaert, L., Boekhoudt, P., Ebbers, W.: Kanaalsturing: Ten gunste van het digitale kanaal. Novay, Enschede (2009) 12. Jaffee, D.: Organization Theory: Tension and change. McGraw-Hill, New York (2001) 13. Jansen, J., Pieterson, W.: Dienstverlening aan bedrijven: Een conceptueel gedragsbeïnvloedingmodel. Universiteit Twente, Enschede (2010) 14. Davis, K.: Human behavior at work: Organizational behavior. McGraw-Hill, New York (1981) 15. Monge, P.R., Contractor, N.S.: Theories of Communication Networks. University Press, Oxford (2003) 16. van Dijk, J.A.G.M.: The Network Society: Social aspects of New Media. Sage, London (2006)

26

Y. van den Boer, W. Pieterson, and R. Arendsen

17. Castells, M.: The New Public Sphere: Global Civil Society, Communication Networks, and Global Governance. The ANNALS of the American Academy of Political and Social Science 616(1), 78–93 (2008) 18. Castells, M.: The rise of the network society. Blackwell Publishing, Cornwall (2000) 19. Janssen, M., Klievink, B.: Improving Governmental Service Delivery: Disintermediation and Re-intermediation strategies for Multi-channel management. In: Proceedings of the 8th European Conference on e-Government, pp. 319–326. Academic Publishing Limited, Reading (2008) 20. Arendsen, R., ter Hedde, M.J.: On the Origin of Intermediary e-Government Services. In: Wimmer, M.A., Scholl, H.J., Janssen, M., Traunmüller, R. (eds.) EGOV 2009. LNCS, vol. 5693, pp. 270–281. Springer, Heidelberg (2009) 21. Pfeffer, J., Salancik, G.R.: The external control of organizations: A resource dependence perspective. Harper & Row, New York (2003) 22. Allers, M.A.: Administrative and Compliance Costs of Taxation and Public Transfers in the Netherlands, dissertation. Rijksuniversiteit Groningen, Groningen (1994) 23. Millard, J.: E-governance and E-participation: Lessons in promoting inclusion and empowerment. In: E-Participation and E-Government: Understanding the Present and Creating the Future (2006) 24. Castells, M.: Communication, Power and Counter-power in the Networks Society. International Journal of Communication 1, 238–266 (2007) 25. Moreira, A.M., Möller, M., Gerhardt, G., Ladner, A.: E-Society and E-Democracy. Paper presented at the eGovernment-Symposium 2009, Berne, Switzerland (2009) 26. Vergeer, M., Hermans, L.: Campaigning on Twitter: Twitter as a campaign tool in the general elections 2010 in the Netherlands. Paper presented at the Etmaal van de Communicatiewetenschap 2011, Enschede, The Netherlands (2011) 27. Breindl, Y.: Critique of the Democratic Potentialities of the Internet: A Review of Current Theory and Practice. TripleC 8(1), 43–59 (2010) 28. Mergel, I.A., Schweik, C.M., Fountain, J.E.: The Transformational Effect of Web 2.0 Technologies on Government (June 1, 2009), Available at SSRN: http://ssrn.com/abstract=1412796 (2009) 29. Daft, R.L., Lengel, R.H.: Information richness: A new approach to managerial behavior and organizational design. In: Cummings, L.L., Staw, B.M. (eds.) Research in Organizational Behavior, vol. 6, pp. 191–233. JAI Press, Homewood (1984) 30. Short, J., Williams, E., Christie, B.: The social psychology of telecommunications. John Wiley, London (1976) 31. Fulk, J., Schmitz, J., Steinfeld, C.W.: A social influence model of technology use. In: Fulk, J., Steinfeld, C.W. (eds.) Organizations and Communication Technology. Sage Pubications, London (1990) 32. Carlson, J.R., Zmud, R.W.: Channel expansion theory and end the experiental nature of media richness perceptions. The Acadamy of Management Journal 42(2), 153–170 (1994) 33. Verhoef, P.C., Neslin, S.A., Vroomen, B.: Browsing versus buying: Determinants of customer search and buy decisions in a multi-channel environment (working paper). Rijkuniversiteit Groningen, Groningen (2005) 34. Postmes, T., Steg, L., Keizer, K.: Groepsnormen en gedrag: Sturing door sociale identiteit en dialoog. In: Tiemeijer, W.L., Thomas, C.A., Prast, H.M. (eds.) De Menselijke Beslisser. Amsterdam University Press, Amsterdam (2009) 35. Hall, R.: Organizations: Structures, processes, and outcomes. Prentice Hall, Englewood Cliffs (1999)

Channel Choice in the G2B Context: Towards a Research Agenda

27

36. Jansen, J., Pieterson, W.: State of the art voor het kanaalkeuzegedrag van bedrijven. Universiteit Twente, Enschede (2010) 37. Nicholson, M., Clarke, I., Blakemore, M.: One brand, three ways to shop: Situational variables and multichannel consumer behavior. The International Review of Retail, Distribution and Consumer Research 12(2), 131–148 (2002) 38. Ebbers, W., Pieterson, W., Noordman, H.: Electronic government: Rethinking channel management strategies. Government Information Quarterly 25(2), 181–201 (2008) 39. de Vos, H.: Kanaalkeuze: Voorkeuren van ondernemers. Novay, Enschede (2008) 40. van den Boom, J.G.G.V.: Ondernemers gaan voor e-overheid: Wat verwacht de ondernemer van de dienstverlening van de overhead?. Ernst & Young, Den Haag (2009) 41. Arendsen, R.: Geen bericht, goed bericht: Een onderzoek naar de effecten van de introductie van elektronisch berichtenverkeer met de overheid op de administratieve lasten van bedrijven. Amsterdam University Press, Amsterdam (2008) 42. Rogers, E.M.: Diffusion of innovations, 5th edn. The Free Press, New York (2003) 43. Wasserman, S., Faust, K.: Social Network Analysis: Methods and Applications. Cambridge University Press, Cambridge (1994)

Supporting Customs Controls by Means of Service-Oriented Auditing Hans Weigand and Faiza Allah Bukhsh Dept. Information Management, Tilburg University, Tilburg, The Netherlands {H.Weigand,f.a.bukhsh}@uvt.nl

Abstract. From a labour-intensive and paper-based door-keeping function Customs control is aiming now towards international trade facilitation in a way that is both efficient and secure. Digitalization of documents and automation of processes plays an important role in this transition. However, the use of modern technology, in particular SOA, is not yet optimal. The research objective of this paper is to explore the potential of SOA for innovating Customs processes by developing a service-oriented auditing approach and evaluating its added-value in the area of Customs control. Service-oriented auditing is a promising tool to implement the ideals of continuous and online monitoring. Keywords: continuous architecture, REA.

1

monitoring,

customs

control,

service-oriented

Introduction

Customs control is an area that is rapidly innovating. From the labor-intensive and paper-based door-keeping function it used to have in the past, it is aiming now towards international trade facilitation in a way that is both efficient to all parties and secure. Digitalization of documents and automation of processes plays an important role in this transition. However, the general feeling is that the use of modern technology, such as Service-Oriented Architecture (SOA) [22], is not yet optimal. Automation of current processes is useful but not sufficient: what is needed is a reengineering of the control processes by going back to the goals to be achieved and the risks to be mitigated. What complicates this problem, compared to the business process reengineering wave of the ‘90s, is that the control processes are interorganizational in nature, involve many parties and are bounded by multiple levels of governmental regulations. The research objective of this paper is to explore the potential of SOA for innovating Customs processes by developing a service-oriented auditing approach and evaluating its added-value in the area of Customs control. The paper is organized as follows. A section 2 reviews recent development in Customs control and introduces the Extended Single Window (ESW) project in which the research is located. This part is based on a literature survey and project meetings we had in 2010 with traders, Customs officers and port community system providers. Section 3 introduces a new T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 28–43, 2011. © IFIP International Federation for Information Processing 2011

Supporting Customs Controls by Means of Service-Oriented Auditing

29

service-oriented auditing approach, which is applied to Customs control in section 4. In terms of Peffer’s design science phases [23], Section 2 contains the identification and motivation, as well as the solution objective of ESW, Section 3 is an exploration of the design and development of the core artifact, and Section 4 is the demonstration. Section 5 closes with conclusions and directions for future research.

2

Customs Control

Customs control is the process where Customs inspects, verifies and examines inward and outward means of transport according to the law to ensure the implementation of national laws and regulations concerning entry and exit of means of transport, goods, personal articles as well as mails and parcels. This is accomplished by declaration procedures, document inspection, examination and other related controls. In 2007, the World Customs Organization developed the so-called International Convention on the simplification and harmonization of Customs procedures (the Revised Kyoto Convention), in an effort to promote the achievement of a highly facilitative international travel and trading environment while maintaining appropriate levels of regulatory control. The Convention explicitly requires administrations to adopt the use of risk management techniques, maximize the use of information technology, implement appropriate international standards and cooperate with other relevant authorities and trade communities, as well as with private sector partners. In this section, we provide an overview of recent developments in Customs control in China and the European Union, and then introduce the Extended Single Window project that started in the Netherlands in 2010. 2.1

China Customs Reforms

China Customs has introduced a series of reforms to improve Customs inspection and supervision. The following description is based on China Customs Annual Report 2009, (http://english.Customs.gov.cn/publish/portal191/). In 2008, China Customs processed 361 million entries and exits of passengers and their personal articles, exerted control over 2.425 billion tons of import & export goods valued at USD 2.56 trillion, strongly accelerating the development of foreign trade as well as cultural and technological exchange. In 2009, China Customs put in efforts to adopt a comprehensive management on the self-compliance of the enterprises, promoting the classified clearance system and implementing the measure of “declare at home, release at port” (a cross-Customs districts clearance model under which qualified enterprises can firstly declare to the Customs house with which they have registered and then have the goods released at the ports where the actual in and outflows of goods take place) as well as the measure of helping goods transfer among different Customs districts. It improves its administration over bonded logistics and makes the clearance processes of Customs special control areas, bonded areas and frontline checking spots gear up to each other.

30

H. Weigand and F.A. Bukhsh

- Customs control over processing trade and bond operations: after getting promise from enterprises that there will be no loss of national revenue, Customs does not levy duties on their import goods but exerts control over the whole process of processing until the processed goods are re-exported. - Audit-based Customs control: within 3 years (or some designated time period) upon release of import or export goods or goods with import duties reduced or exempted, Customs verifies the account books, related documents and relevant import & export goods of the legal entities under Customs control, so as to supervise the truthfulness and legitimacy of their import & export activities. The priority of audit-based control is to regulate the importation and exportation of the enterprises, guide them towards compliance and self-discipline and find smuggling or violation clues through post auditing on goods of normal trade, and auditing on goods whose duty reduction or exemption has been approved. - Customs statistics: the process where Customs conducts statistics compilation and comprehensive analysis. Attributes include description, quantity, weight, price, country (region) of origin, legal entity, domestic destination, domestic place of origin, trade method, transport means and relevant Customs districts (houses) of the import and export goods. The objective of these statistics is to reflect comprehensively and accurately the faring tendency of foreign trade, conduct early-warning import & export monitoring and provide information services for national macro-economic decision-making and the development of foreign trade. 2.2

EC Modernized Customs Code

The Modernized Customs Code (MCC) was adopted by the European Community in April 2008 but the process of implementation is still continuing. The MCC will simplify legislation and administration procedures both from the point of view of Customs authorities and traders. It aims to: - Streamline and reduce the number of Customs procedures and make it easier to keep track of goods; - Ensure the progressive automation of all Customs formalities, with a view to a completely ‘paperless environment for Customs and trade’. This applies to documents between traders and Customs as well as between Customs and other authorities. - Promote the concept of ‘centralized clearance’, under which authorized traders will be able to declare goods electronically and pay their Customs duties at the place where they are established, irrespective of the Member State through which the goods will be brought in or out of the EU Customs territory or in which they will be consumed. - Offer bases for the development of the ‘Single Window’ and ‘One-Stop-Shop’ concepts, under which economic operators give information on goods to only one contact point, (Single Window concept) even if the data should reach different administrations/agencies, so that controls on them for various purposes (Customs, security, sanitary,...) are performed at the same time and at the same place (One-stopshop concept). - Streamline and harmonize further the Customs guarantee systems.

Supporting Customs Controls by Means of Service-Oriented Auditing

31

The concept of centralized clearance means that when an "authorized operator" lodges his Customs declaration at the Customs office where he is established, that office would then carry out any documentary risk analysis. Subsequently, this office would forward the results of its analysis to the border Customs office in that Member State or in another Member State where the goods are actually to enter or leave the Community. This border office would apply any physical controls to the goods being imported or exported that either office deems necessary on the basis of the result of risk analysis. Compliant and trusted traders would generally have a minimal incidence of selection for control at the place of entry or exit. Provided that there is no irregularity, and that the Customs debt was provided for at the office of import, the goods would be released directly into free circulation in the EU or for export. Under the procedure, the goods do not have to be moved to the office of import or export but could be delivered directly to the point of sale, in whatever country. So this allows multi-national companies to conduct all of their EU business with one Customs office. 2.3

Extended Single Window

The vision of the Extended Single Window project (ESW) that started in 2010 is to develop an integrated coordinated border management solution for ports and airports integrating with previous and subsequent procedures for reliable, secure, and cost effective logistic chains throughout the Netherlands as a logistic gateway to Europe. This coordinated border management solution is referred to as ‘Extended Single Window’. It requires efficient and reliable handling of data to generate information for effective joint supply chain planning for shippers, goods owners, transportation companies, forwarders, terminals and other logistic service providers and to use these data also to generate information for government agencies, like Customs, agricultural and tax. Currently, shippers and goods owners are faced with a wide range of regulations and procedures when goods enter or exit the EU (border, agricultural, safety, import controls, statistics, and indirect taxes). Completion of declaration processes and risk analyses and planning and coordination of inspections by the various agencies before shipments are (un)loaded from an aircraft or vessel enables logistics actors (terminal operators, forwarders, transport operators) to plan and execute transportation of shipments with hinterland hubs efficiently (improving modal shift, throughput time (i.e. for perishable goods) and reducing congestion). Efficient and reliable government controls reduce administrative costs, increase reliability of the supply chain, and ultimately reduce transport costs for shippers and logistic operators. Thus, ESW aims to contribute to the implementation of the MCC for both incoming and outgoing logistic flows, including integration with previous (outgoing goods for instance preceded by export) and subsequent procedures (incoming goods for instance followed by transit). Basic research in advanced information technologies is in Event Driven Architecture with a Logistic Interoperability Ontology: -

Event Driven Information Service Bus (EISB). This is a (virtual) technical architecture at middleware level that extends the concept of Enterprise Service Bus (ESB) as it is advocated in SOA. Basically, each logistic operation triggers an event. Discharge of a container from a vessel and loading cargo on a truck for

32

-

H. Weigand and F.A. Bukhsh

on-carriage are examples of events. Minimally, an EISB supports publish/subscribe functionality to events, thus supporting a virtual data space. Since the data space is virtual, relevant data can still reside with each actor depending on governance and logistic innovations at business level. Authorities can subscribe to events upon which processes can be triggered for retrieving data required by risk management systems or coordinating inspection. The EISB fully supports a SOA since service requests and responses are treated as events as well. Conversely, subscribing to a published event type is equivalent to using the event publisher as a service (typically called monitoring service or sensor service). In this way, the EISB can support traditional document-driven processes as well as new event-driven processes based on smart seals (RFID etc.) for tracking and tracing of movement of goods. Logistic Interoperability Ontology Framework. It specifies the semantics of all physical objects as shared by business actors in supply chains, e.g. semantics of containers, goods items, and trucks. The ontology also encompasses mechanisms for information sharing amongst all actors in logistic chains thus allowing that each actor shares only relevant information with one or more other actors.

Using the EISB concept it is possible to extend the Single Window concept in at least two significant ways. Firstly, the Single Window is based on digital documents, whereas the ESW is based on events, which is much more flexible. Assume that a sender (shipper) has to send a document to a number of receivers (Customs office, Tax office). He fills in the data in a document template. With Single Window, he has to send the document only once, but filling in the document is still mainly a process of copying data from one source to another. With Extended Single Window, the data collection and sending may become superfluous. Events are published only once. The receiver specifies which data he wants to see (by subscribing to events), and these data are collected then (that is, continuously) from the virtual data space fed by all the distributed events. Secondly, the Single Window only streamlines data flow in one direction, from logistic operators to government agencies, whereas the EISB also supports data flow among logistic operators and across the supply chain, among government agencies (e.g. to realize a One-Stop-Shop), and from government agencies to operators. One of the very powerful new possibilities opened up this way is end-to-end supply chain integrity as advocated by [10]. Whether such an EISB is managed by a single intermediary, or that the implementation involves an intermediary network with an industrial part and a governmental part, is still an open issue, but has no effect on the EISB concept as such. Building on the EISB, the ESW project furthermore aims to reengineer Customs procedures using service-oriented technology. In the envisioned scenario, Customs can use audit services that monitor logistic flows via the EISB, preferably in a continuous and non-intrusive way. In the next section, we will develop a framework for Service-Oriented Auditing. The application of this solution approach to Customs requirements is demonstrated in section 5.

Supporting Customs Controls by Means of Service-Oriented Auditing

3

33

Service-Oriented Auditing

The primary objective of a traditional audit of an organization’s financial statements is to form an opinion on the trustworthiness of the included information and to make this opinion public to an interested audience. Such an opinion is based on a careful analysis of the potential risks and the effectiveness of built-in control mechanisms to mitigate them, - influencing correctness and completeness - as well as a check on the company-wide integrity of the information. An auditor will look for assurance by either auditing the design, i.e., the implementation and effectiveness of controls, or, alternatively, by looking substantively at the data generated by the system. The latter approach can be very costly. These costs can be reduced by either taking a random sampling approach or, nowadays, by modern process mining techniques such as described in [1]. Of course, in practice always a combination of the two approaches will be used, as it is now, but the weight assigned to each can shift significantly. 3.1

Auditing and ICT

As modern information systems draw heavily on ICT, modern auditing cannot do without ICT either. Many tools to support the auditor are already available, e.g. ACL and IDEA. Still, we have identified many research challenges in this field: • Semantic gap. According to Li [16], there is quite a gap between the IS (Information Systems) environment and the audit environment. It is not sufficient for auditors to have direct access to databases and files; they should also understand the semantics. The same applies to processes. An analysis of business modeling approaches from the perspective of process level audit risk assessment by Carnaghan [4] reveals that current approaches do have their merits but also their lacks. • Managing the business logic. The interpretation steps made during the monitoring process are warranted by business rules. These business rules (the policies they capture) change over time, and therefore need to be managed properly. This change management is complicated by the above-mentioned semantic gap that exists between the execution level addressed by the IT and the policy level on which the business formulates its requirements. • Smart auditing. There are more and more AI techniques that can be used to mine large amounts of data (off-line or real-time), such as neural networks, genetic algorithms, process mining and Bayesian belief networks. In addition, the intelligent use of domain knowledge can often simplify or optimize heavy data search processes [9]. • Adaptive monitoring. Monitoring can be a complex task that uses a configuration of numerous hardware and software resources. The question is not only whether it can be done but also how to optimize this task (in terms of time, costs on the one hand and the effectiveness in terms of risk management and control on the others), and how to cope with failures. The aim of adaptive monitoring is to allow dynamic configuration of the monitoring task according to certain performance goals.

34

H. Weigand and F.A. Bukhsh

• Online auditing. Online (continuous) auditing means that the auditor does not inspect the organization physically, but rather inspects databases remotely. This format has certain limitations, but also new possibilities, in particular (a) the possibility to run complex queries and (b) the possibility to cross-check with other databases (public databases or business partner databases) and (c) to do this on a real-time continuous basis [14]. Online auditing should be non-intrusive, i.e., not interrupt normal operations, and obey transparent security requirements. Online auditing is closely related to the idea of “Piggy-Backing” where auditor and auditee agree to use the same business data flow for both internal and external audit purposes. Continuous Online Monitoring started at the end of the 90’s to draw on Embedded Audit Modules. Embedded Audit Modules (EAMs) are software applications embedded in host systems or linked to host systems in order to externally monitor such systems. EAMs continuously monitor flows of transactions, identify transactions that match certain pre-determined integrity constraints and, in the event of a constraint violation, alert the auditor and copy the transaction data to a file. The adoption of EAMs has been rather low so far [6]. As already argued by [21], web service technology can make the implementation much simpler, and also broaden the application from single enterprise ERP systems to supply chains, and from single databases to event logs and business process audit trails. Conceptually, a distinction must be made between Continuous Monitoring and Continuous Auditing. The American Institute of Certified Public Accountants (AICPA) defines continuous auditing as „written assurance that is provided at the time or very near the time of events that affect the financial statements‰. In contrast, continuous monitoring does not necessarily involve assurance, but it can be used to realize the former. In the following, we will focus on two of the above-mentioned challenges, while considering the others as well. To address the semantic gap, we propose to base the auditing on the REA business ontology. To realize Continuous Online Monitoring, we propose a SOA-based solution. 3.2

REA Business Ontology

The Resource-Event-Agent (REA) ontology was first formulated in [18]. The following is a short overview of the core concepts of the REA ontology. A resource is any object that is under the control of an agent and regarded as valuable by some agent. This includes goods and services. Resources are modified or exchanged in processes. A conversion process uses some input resources to produce new or modify existing resources, like in manufacturing. An exchange process occurs as two agents exchange (provide, receive) resources. To acquire a resource an agent has to give up some other resource. An agent is an individual or organization capable of having control over economic resources, and transferring or receiving the control to or from other agents. The constituents of processes are called economic events. REA recognizes two kinds of duality between events: conversion duality and exchange duality.

Supporting Customs Controls by Means of Service-Oriented Auditing

35

The reason for building Service-Oriented Auditing on REA is that REA abstracts from process details and implementation systems, and so talks the language of the business rather than the IT specialist. This is because it focuses on economic value. At the same time, it has been shown that REA structures also provide a solid basis for implementation. There is a direct mapping from REA models to database structures [5]. The dualities express integrity constraints that that can be used for both the design of control mechanisms (preventive) and for the detection of deviating behavior (detective) that may indicate fraud. From an advanced audit perspective, it is not always sufficient to consider operational processes only. It may also be necessary or more efficient to include the management level. That is why the policy level of REA must be taken into account as well. The event records give an answer to the question “what has happened?”, but not to the question “what is planned or scheduled – what should happen?” The latter is modeled at the policy level that allows talking about types and groups as well as commitments. Policies can be formalized in analogy to contracts as a group of intentional resources obeying the duality principle: resources that the agent gives in versus what he aims to achieve [27]. On the event dimension, constraints are what the agent gives in (directives that limit the actions of the controlled system, e.g. an authorization table) and goals are what the agent aims to gain in return for that (the expected result of the controlled system, in terms of evaluations). In addition, the policy may contain assumptions in the form of testable assertions. The constraint part of the policy is the basis for (decision services within) the operational process design, the goal part is the basis for the service monitoring design, and the assumption part is the basis for the context monitoring design. 3.3

Continuous Monitoring

In order to audit the core enterprise, the auditor must identify the Soll (de jure) and 1st (de facto) modality (cf. [1]). Buffer contents such as inventories or bank accounts may have illicit decreases or increases. Such errors have consequences for the other audit objects. For instance, an illicit decrease may lead to a potential overstatement of the registration, and hence of the financial statements (although possibly not material). We define the process structure that abstracts from such errors to be the Soll modality, whereas the 1st modality is obtained by incorporating the identified potential error classes. The 1st model can help to design internal controls or analyze their effectiveness in either preventing illicit events (preventive internal control) or being able to notice them when they occur (detective internal control), including being able to notice violations of the internal controls themselves (compliance procedures). The REA event model, containing all the relevant economic events, has 1st modality. It does not incorporate all error causes, and so it is not a complete event model. However, as the REA model contains all resources that matter for the company, it has, by definition, a handle on all error effects. The REA model in terms of its duality constraints has the Soll modality. More normative constraints, such as permissions and authorizations, can be specified on the REA policy level [8]. We use “process patterns” for representing both the normative process structure and the potential deviating ones.

36

H. Weigand and F.A. Bukhsh

Each process pattern is specified concretely in two layers: the REA layer (REA events), and the implementation layer (events such as generated and available in the operational system). When the pattern matches a certain trace of operational events, then the REA layer of that pattern interprets this trace in REA terms. The normative process structure is a combination (“weaving”) of the core process with built-in controls (cf. [7,15]). Built-in controls can take many forms, including segregation of duties, restricted authorizations, required recordings (event traces), and pre-conditional checks. After we have identified different control strategies, they can be evaluated on the basis of effectiveness (as defined above) and efficiency (costs).

Fig. 1. Adaptive continuous monitoring service. The filled boxes represent activities, the dotted boxes informational resources.

Fig. 1 depicts the overall architecture of an adaptive service-oriented monitoring solution. Business services generate events that are published on the EISB. The continuous monitoring (CM) service also taps into the EISB, locally or remotely (as in [12]). The CM service collects the events using the publish/subscribe mechanism. Then it generates the IST model from the event traces by means of the specified process patterns. When the processing detects an operational event pattern, it generates the corresponding economic event as IST model. The IST model contains both good and bad process instances from which a fault list is generated (evaluation). This information can be forwarded immediately to the responsible actor, or can be logged on permanent storage for later inspection purposes (publication). The fault list can also be sent to the management control service of the process being monitored so that it can take appropriate measures (this does not hamper the independence of the monitoring/auditing service as long as the monitoring service is not involved in the control itself). Monitoring management is the activity (manual, semi-automated or perhaps fully automated) that starts with deriving process patterns from a given SOLL model. The monitoring management turns the monitoring into an adaptive process by adapting and extending the process patterns on the basis of the feedback that it gets

Supporting Customs Controls by Means of Service-Oriented Auditing

37

from the processing. Adaptations can be in the form of new patterns actively monitored or in adapting existing patterns, e.g. to make the collecting more efficient. Currently, the adaptation should be seen as a manual process (but see below). The CM service such as depicted in Fig. 1 is first of all a tool that works on the process instance (data) level. However, type level (i.e., design) evaluation can be implemented using the same service architecture, but with different process patterns. The type level REA layer consists basically of (quantified) duality axioms and other “continuity equations” [25] that relate aggregated variables (e.g. total “cash inflow” or “inventory level”). These continuity equations don’t need to be specified manually but are derived automatically from the SOLL REA model. Evidently, type level control patterns are evaluated not continuously, but periodically, although the period does not need to be as long as a year or quarter. At the moment, there are several techniques for process mining available, and powerful SQL-like Event Processing Languages that can be used to express process pattern conditions (e.g. [26]). Still, there are several opportunities to extend the CM model. We describe what we see as the three most important ones: 1.

2.

3.

3.4

The CM service can be strengthened by including a validation step. Research studies in the practice of embedded audit modules [2] found that they tend to give a huge list of false positives. As this may threaten the usability, it is better to make a distinction between process patterns that identify a certain event and process patterns that can be used to validate it, using independent evidence. Depending on the application, the evaluation can be performed on validated IST models only. EAM modules and event-driven service busses usually take an event trace to be a record of structured numeric data. However, it is possible to widen the scope and include communication events such as emails, discussion groups, social media (Facebook) updates etc, and mine the text content as well. Text mining is still in an experimental phase, but there are already some results (e.g. [17]). The basic CM service described here already supports what Hunton and Rose [13] call dynamic auditing, where the configuration of the monitoring service is not fixed but adaptive. The monitoring management service implements a diagnostic control loop, but with manual adaptation. The next step is self-adaptation. According to [27] a self-adaptive management service should be built as a rulebased system, where rules do not only specify under which conditions which adaptations have to be made, but also the assumptions behind the rule. The assumptions are continuously checked by a special management subservice, typically using external sources, e.g. on economic and political news. For instance, in the case of political unrest in the country of origin of a certain cargo, the standard risk assessment rule may have to be replaced by a stricter variant. System-Based Auditing

System-based auditing is already known since the ‘70s of last century. Basically, it means that the auditee shows that he is “in control” of the processes in question. Being in control means that the process is guaranteed to comply with the control

38

H. Weigand and F.A. Bukhsh

norms (SOLL model) and/or that all deviations are known and addressed. This implies that the auditee has a full specification of the process including built-in controls and a complete fault list. When the process is fully automated, it is not too hard to get assurance that such a full specification is also executed, when the correct functioning of the software has been assured by IT auditing. In a service-oriented system-based audit, an independent audit service continuously monitors both the auditee’s process specification and the fault list upon changes. In an extended form, the monitoring also includes the governance process of reacting to deviations (the follow-up of the fault list) and adapting the specifications, according to a SOLL model of the management process. In that respect, it helps when the management is implemented as a semi-automated service itself, like the CM service above. So in that case, the auditee has a CM service as in Fig. 1, including the monitoring management, and the auditor (say, Customs) has a CM service that monitors the first one.

4

Service-Oriented Customs Control

This section draws upon the Service-Oriented Auditing approach of section 3 to identify innovative web services for Customs control. We start with a top-down analysis of the Customs services and their requirements as collected in section 2.

Fig. 2. Customs as a Service – global view

The main services identified are Security, Tax/Excise and Statistics. The value of a security service is that it aims at mitigating security risks, where we use Keeney’s widely cited definition of risk as “the possibility that human actions or events lead to consequences that harm aspects of things that human beings value”. In other words, a security service creates value by preventing the value of certain resources to be diminished. Tax and Excise services create monetary value by collecting payable duties. Customs statistics service creates value in the form of useful information,

Supporting Customs Controls by Means of Service-Oriented Auditing

39

either directly to interested actors or via other governmental statistics offices. All these services are somehow related to cross-border supply chains, in which we can distinguish many business events, such as shipment, transport and payment. The three main Customs services are complex services that make use of a limited number of generic core services. So far, we have identified five core services: risk detection, risk mitigation, liability detection, liability acquittal, and information (Fig. 2). An information service provides aggregated data based on all the events that it monitors. So far, this information takes the form of standard regular reports. In the future, service-oriented solutions will be possible in which the data don’t need to be pushed, but can be pulled anytime by any (authorized) interested actor, according to user-defined queries (with a similar structure as the process patterns described in 3.3), and including drill-down functions to higher levels of detail. Customs tax and excises are raised according to law-governed liabilities. A major distinction can be made between the detection or creation of liabilities and their acquittal. We will come back to these below. A risk detection service is a Continuous Monitoring service such as defined above where the SOLL model is based on values to be protected, and the IST model contains a context model of all relevant events. A risk mitigation service takes measures to mitigate risks, e.g. by inspections or interventions.

Fig. 3. REA model of excise liability life cycle

4.1

Excise Management

As an example of liability detection and acquittal, we consider the case of excise management. Excise taxes are classified by the OECD as being those taxes which are “levied on particular products, or on a limited range of products … imposed at any stage of production or distribution and are usually assessed by reference to weight or strength or quantity of the product, but sometimes by reference to the value”. In his

40

H. Weigand and F.A. Bukhsh

description of key controls in excise administration, Preece [24] distinguishes between liability creation events and liability acquittal events. Liability is created not only by import, but may also be created by returns of deliveries, manufacturing, gains during storage or manufacturing, and bonded receipts. Liability is acquitted not only by payment, but may also be acquitted by export, bonded sale, manufacturing into new products, detoriation and losses during storage or manufacturing. Fig. 3 provides a REA model of the liability life cycle. At the top-right, we have modeled the fundamental exchange duality between liability acquittal (in the form of a charge) and payment. This duality is a typical example of many e-government interactions with business. The acquittal process itself – including standard legal provisions for objections etc. – and the payment handling follow standard business process patterns, and it is seems very well possible that not the Customs agency itself, but a central government service takes care of this part. This does not apply to the other ways in which an excise liability can be acquitted. The pièce de résistance is the creation of liability. In the model we have represented this as the “excising” event. Essentially, levying an excise draws on excise law and is attached to products (rather than to events), but triggered by events such as the production or importing. In an EISB approach, there are different ways to implement the excising event. Customs may install a rule-based decision service that monitors the EISB for the relevant events, such as “importing” in order to generate excise events that can be picked up by the acquittal service. Alternatively, companies with AEO status can include such a decision service in their own system, and Customs only needs to perform a system-based audit to check whether it is configured correctly. From an audit perspective, it is useful and often necessary to consider the generating events in a broader context. In the case of production, there is a conversion duality with the material used in the industrial production process. Importing is part of a sales process, which can be used either as independent evidence of primary source. Sales have an exchange duality with the payment, yet another independent evidence. 4.2

Security Management

Shortly after 9/11, the US Customs Container Security Initiative (CSI) and Customs Trade Partnership Against Terrorism (C-TPAT) were announced. The philosophy of C-TPAT is two-sided: on the one hand, cooperate with international trade actors to improve the security of their supply chains; and on the other hand to focus efforts and resources on potentially high-risk consignments that do not form part of such a secured supply chain. The new instruments available to security management are manifold. A first distinction can be made between risk mitigation services and risk detection. Risk detection – using the CM model – can be split up in collecting, processing and evaluation. For collecting, a whole range of new technologies are upcoming, such as smart containers, high-energy scans, low-energy scans and GPS. Using an EISB, the monitored events can be published over the whole supply chain (although some optimization of the data distribution will be a necessary feature of the bus). As argued by Hesketh [11], current Customs “fail to recognize the strategic importance of

Supporting Customs Controls by Means of Service-Oriented Auditing

41

accurate data at the point of consignment completion and export. An importer is not best placed to make a legal declaration about goods packed by the consignor”. To increase safety, it is absolutely necessary to have information that can only be provided by the consignor who “packed the box”. At the processing level, it is possible to use crime script analysis and artificial intelligence techniques [19]. Risk evaluation is based on risk norms. In principle these norms are derived from governmental regulations, but at the operational level it is necessary to make use of built-up experience. For the first part, it is possible to make use of legal document management services such as described in [3] for the latter the use of machine learning techniques is becoming indispensable. The risk detection service is to be developed by each country in isolation, but such a service can greatly benefit from being integrated in an international network that exchanges risk events, crime scripts and some of the learned risk norms in a swarm intelligence style [20].

5

Conclusions

Customs control is in a massive transition process of innovation and reorganization. In this paper, we have shown that a service-oriented architecture can foster this transition by providing a flexible and adaptive integration platform. We have introduced a service-oriented auditing module that combines REA business modeling with adaptive Continuous Auditing techniques. Its application to Customs control looks promising. It allows for a loose integration of software services across the supply chain, between traders and government, and among governmental agencies. The service-oriented auditing module is a starting point only. It is necessary to substantiate it further by actually selecting and combining the techniques mentioned above into real audit services and by developing standardized interfaces. As far as REA is concerned, more research is needed on extending its applicability from a single company to a business network or chain. Another topic for future research is the development of an analytic tool to analyze the effectiveness and efficiency of monitoring and auditing configurations in a sound and systematic way. Acknowledgements. This research is supported partially by the DINALOG project Extended Single Window (www.dinalog.nl).

References 1. van der Aalst, W.M.P., van Hee, K., van der Werf, J., Verdonk, M.: Auditing 2.0: Using process mining to support tomorrow’s auditor. Computer 43(3), 90–93 (2010) 2. Alles, M., Kogan, A., Vasarhelyi, M.: Putting Continuous Auditing Theory Into Practice: Lessons from Two Pilot Implementations. Journal of Information Systems 22(2), 195–214 (2008)

42

H. Weigand and F.A. Bukhsh

3. Boer, A., van Engers, T., Winkels, R.: Traceability and Change in Legal Requirements Engineering. In: Casanovas, P., Pagallo, U., Sartor, G., Ajani, G., et al. (eds.) AICOLII/JURIX 2009. LNCS, vol. 6237, pp. 74–92. Springer, Heidelberg (2010) 4. Carnaghan, C.: Business Process Modeling approaches in the context of process level audit risk assessment: an analysis and comparison. Int. Journal of Accounting Information Systems 7, 170–204 (2006) 5. Chang, C.J., Ingraham, L.R.: Modeling and Designing Accounting Systems: Using Access to Build a Database. Wiley (2007) 6. Debreceny, R.S., et al.: Embedded Audit Modules in Enterprise Resource Planning Systems: Implementation and Functionality. Journal of Information Systems 19(2), 7–27 (2005) 7. Gal, G., McCarthy, W.: Specification of internal accounting controls in a database environment. Computers & Security 4(1), 23–32 (1985) 8. Gal, G., Geerts, G., McCarthy, W.: Semantic Specification and Automated Enforcement of Internal Control Procedures within Accounting Systems. Presentation at 3rd VMBO Workshop (Value Modeling and Business Ontologies), Stockholm (2009) 9. Graco, W., Semenova, T., Dubossarsky, E.: Toward knowledge-driven data mining. In: Proc. of the 2007 Int. Workshop on Domain Driven Data Mining (DDDM 2007). ACM, New York (2007) 10. Hesketh, D.: Seamless electronic data and logistics pipelines shift focus from import declarations to start of commercial transaction. World Customs Journal 3(1), 27–32 (2009) 11. Hesketh, D.: Weaknesses in the supply chain: who packed the box? World Customs Journal 4(2), 3–20 (2010) 12. Holmes, T., Zdun, U., Daniel, F., Dustdar, S.: Monitoring and Analyzing Service-Based Internet Systems through a Model-Aware Service Environment. In: Pernici, B. (ed.) CAiSE 2010. LNCS, vol. 6051, pp. 98–112. Springer, Heidelberg (2010) 13. Hunton, J.E., Rose, J.: 21st Century Auditing: Advancing Decision Support Systems to Achieve Continuous Auditing. Accounting Horizons 24(2), 297–312 (2010) 14. Kogan, A., Sudit, E., Vasarhelyi, M.: Continuous online auditing: a program of research. Journal of Information Systems 13, 87–103 (1999) 15. Lee, R.M., Bons, R.W.H., Wagenaar, R.W.: Pattern-directed Auditing of Interorganisational Trade Procedures. In: Schmid, B., et al. (eds.) Towards the e-Society: First IFIP Conference on E-Commerce, E-Business, and E-Government, I3E 2001. Kluwer Academic Publishers (2001) 16. Li, S.H., Huang, S.M., Lin, Y.C.: Developing a continuous auditing assistance system based on information process models. Journal of Computer Information Systems (Fall 2007) 17. Ma, Z., Sheng, O., Plant, G.: Discovering company revenue relations from news: A network approach. Decision Support Systems 47, 408–414 (2009) 18. McCarthy, W.E.: The REA Accounting Model: A Generalized Framework for Accounting Systems in a Shared Data Environment. The Accounting Review (1982) 19. Hermanson, D.R., Moran, B., Rossie, C., Wolfe, D.: Continuous Monitoring of Transactions to Reduce Fraud, Misuse and Errors. Journal of Forensic Accounting, 17–30 (2006) 20. Kennedy, J.R., Eberhart, C., Shi, Y.: Swarm Intelligence. Morgan Kaufmann (2001) 21. Murthy, U., Groomer, S.: A continuous auditing web services model for XML-based accounting systems. Int. J. of Accounting Information Systems 5(2), 139–163 (2003) 22. Papazoglou, M.P., van den Heuvel, W.J.: Service oriented architectures: approaches, technologies and research issues. VLDB Journal 16(3), 389–415 (2007)

Supporting Customs Controls by Means of Service-Oriented Auditing

43

23. Peffers, K., Tuunanen, T., Rothenberger, M., Chatterjee, S.: A Design Science Research Methodology for Information Systems Research. Journal of Management Information Systems 24(3), 45–77 (2008) 24. Preece, R.: Key Controls in the Administration of Excise Duties. World Customs Journal 2(1), 73–92 (2008) 25. Vasarhelyi, M., Alles, M., Kogan, A.: Principles of analytic monitoring for continuous assurance. J. Emerg. Technol. Account. 1(1), 1–21 (2004) 26. Wu, E., Diao, Y., Rizvi, S.: High-performance complex event processing over streams. In: Proc. of the 2006 ACM SIGMOD Int. Conf. on Management of Data (SIGMOD 2006), pp. 407–418. ACM, New York (2006) 27. Weigand, H., Johannesson, P., Andersson, B., Jayasinghe Arachchige, J., Bergholtz, M.: Management Services – A Framework for Design. In: Mouratidis, H., Rolland, C. (eds.) CAiSE 2011. LNCS, vol. 6741, pp. 582–596. Springer, Heidelberg (2011)

IMP2.0 Migration Collaborative Platform and e-Services Nineta Polemi, George Pentafronimos, and Theodoros Ntouskas University of Piraeus, Dept. of Informatics, Karaoli & Dimitriou 80, 18534, Pireaus, Greece {dpolemi,gpentas,tdouskas}@unipi.gr

Abstract. Legal immigration is a complex, multi dimensional issue and the European Commission (E.C.) project ImmigrationPolicy2.0 (IMP2.0) [1] considered it as such by studying its legal, societal and policy implications before proposing and implementing holistic e-migration services contributing towards participatory governance and the harmonization of existing policies and practices. This paper outlines the IMP2.0 social, policy and legal research results which led to the design and implementation of targeted e-migration collaborative services by presenting the identified requirements and describing the services provided by the IMP2.0 web platform. Keywords: legal immigration, collaborative technologies, e-migration services.

1

Introduction

IMP2.0 [1] is a technological project driven from the needs arising in the migration policy field. Social Integration, Social Cohesion, Security, Privacy and Interoperability are the most important European Union (E.U.) policy objectives that IMP2.0 directly addresses. It adopts a bottom up approach by analyzing all aspects (legal, societal, policy) of the complex migration employment problem (the most important part of a Social Integration Policy) and provides e-services which contribute towards harmonizing the efforts for establishing a common E.U. migration employment policy. It provides an innovative IT instrument to both migration related communities (policy makers and legal immigrants), using new technologies and opens new research areas (visualization, collaborative modeling and graphics, governmental process management systems, collaborative decision making). In contrast to the present situation of IT governmental tools, IMP2.0 paves the way for a new research area in “Innovative migration electronic systems”. In this paper we outline the IMP2.0 social, policy and legal research results which led to the design and implementation of targeted e-migration collaborative services and it is organized as follows: In Section 2, the E.U. policy and legislation harmonization efforts related to employment migration are outlined and open problems and gaps at practical level are identified. In Section 3, the user requirements as reported in the IMP2.0 survey involving four hundred and sixty eight (468) stakeholders (policy makers and legal immigrants) in the migration employment T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 44–57, 2011. © IFIP International Federation for Information Processing 2011

IMP2.0 Migration Collaborative Platform and e-Services

45

community are presented. Considering these requirements, IMP2.0 designed and implemented targeted e-migration services, in order to contribute effectively towards the efforts in harmonizing E.U. migration employment policies as described in Section 4. The paper concludes with Section 5 in which the main results of the paper are outlined and further research propositions are provided.

2

Migration Employment Policies and Legislation

2.1

European Union Efforts

IMP2.0 performed a detailed desk study on migration policies at national and European level. In this section the main results of the European level desk study will be presented (for detailed analysis the reader is referred to [3]). The European Union (EU) Council acknowledges that migration employment issue, by its very nature, can be better addressed at European rather than at national level. It emphasizes the importance in harmonizing migration policies [10] starting its efforts towards this direction from 1999 with its three main policy Programmes, Tampere, Hague and Stockholm, [4]: The Tampere program (1999-2004) in which the E.U. leaders at the 1999 E.U. Council in Tampere (Finland) agreed upon that a comprehensive approach to migration addressing political, human rights and development issues in countries and regions of origin and transit is needed. This requires improving living conditions and job opportunities, preventing conflicts and consolidating democratic states and ensuring respect for human rights, in particular rights of minorities (women and children). Partnership with third countries concerned is also a key element for the success of such a policy, with a view to promoting cooperative development. From a legislative point of view, the main achievements during the period of implementation of the Tampere program have been the following: • Family reunification – The Council Directive 2003/86/EC [14] on the right to family reunification. Member States’ legislation had to comply with this Directive not later than 3 October 2005. • E.U. long-term resident status - The Council Directive 2003/109/EC [13] on a long-term resident status for third country nationals who have legally resided for five years in the territory of a Member State. Member States legislation had to comply with this Directive by 23 January 2006. • Students - A Directive on the conditions of admission of third-country nationals for the purposes of studies, pupil exchange, unremunerated training or voluntary service was adopted by the Council Directive 2004/114. Member States’ legislation must comply with the Directive by 12 January 2007. • Researchers – A Directive for the facilitation of the admission of researchers into the E.U. was adopted by the Council Directive 2005/71. The Hague program (2005-2010) sets the objectives for strengthening freedom, security and justice in the E.U. for this period. The 2005 E.U. Council, taking into account the outcome of discussions on the Green Paper on an E.U. approach to

46

N. Polemi, G. Pentafronimos, and T. Ntouskas

managing economic migration (COM (2004) 811) addressing labor migration, best practices in Member States and its relevance for implementation of the Lisbon strategy, invited the Commission to present a Policy Plan on Legal Migration (COM (2005) 669). The document foreshadowed the proposal for a directive on a single application procedure for a single permit for nationals of third countries to reside and work lawfully in the territory of a Member State and on a common set of rights for third-country workers legally residing in a Member State, which was adopted by the Commission in October 2007 (COM (2007) 637 final). In September 2005, the Commission adopted the communication 'A Common Agenda for Integration Framework for the Integration of Third-Country Nationals in the European Union' (COM (2005) 389).This Communication provides new suggestions for harmonization action both at E.U. and national level. In December 2005, the E.U. Council adopted the Global Approach to Migration, and considers migration issues as integral part of the E.U.'s external relations. In June 2006 the Commission presented the Second Annual Report on Migration and Integration [SEC (2006) 892] which provides an overview of migration trends in the E.U. In May 2007, the second edition of the 'Handbook on Integration for policymakers and practitioners' was issued focusing on mainstreaming immigrant integration, housing in an urban environment, economic integration and integration governance. In September 2007, the Third Annual Report on Migration and Integration (COM (2007) 512) depicts the continued monitoring process of harmonized policy developments on admission and integration of third-country nationals in the EU. The 2008 E.U. Council endorsed the Global approach to migration. Its purpose is, to create a coherent policy approach in the area of migration: Development policy, measures to encourage legal migration and fight illegal immigration, as well as managing demand for skilled labor in a framework of dialogue, cooperation and partnership with countries of origin and transit. In 2009, the Commission adopted a directive on the admission of highly-skilled third country nationals, the so-called EU-Blue Card, a fast-track procedure for issuing a special residence and work permit under more attractive conditions. It is the only directive derived from the Policy Plan on Legal Migration from 2005 adopted to date. The Stockholm Program (2010-2014) approach differs from its predecessors (Tampere Program and Hague Program) with regard to several aspects. For example, the Hague Program emphasized harmonizing migration and asylum policies across Member States, where the Stockholm Program prioritizes initiatives promoting cooperation between the Member States to find practical solutions [11]. It also seeks to use soft law such as guidelines and stand-alone E.U. agencies, rather than the hard law envisioned in the Maastricht Treaty. The program sets out the following priorities in the area of immigration, in which the relevance of labor migration is curtailed compared to the Hague Program: • Global Approach to Migration: The need to identify common interests with third countries is highlighted. • Migration and Development: Facilitating remittances, cooperation with diaspora groups, and promoting circular migration constitute EU’s priorities in the migration development nexus.

IMP2.0 Migration Collaborative Platform and e-Services

47

• Labor Migration: The need for improved information and data on skills needs, skills recognition and development as well as skills matching with third countries were highly considered for further actions. • Rights for third-country nationals: Third-country nationals shall be endowed with "rights and obligations comparable to those of E.U. citizens" by the end of the program’s period. • Integration: Within the program several ideas in the area integration are proposed, e.g. the development of common European modules, as well as E.U.-wide indicators to monitor the results of harmonized integration policies. • Illegal Migration: The need for a sustainable return policy supplemented by encouragement of voluntary return, and renewed efforts to sign readmission agreements with more countries of origin and transit. • Unaccompanied Minors: Unaccompanied minors, not addressed in the previous programs, are included in the Stockholm Program. In April 2010 the Commission adopted an action plan on the Stockholm Program (COM (2010) 171 final) to implement the political priorities set out in the program. The third edition of the Handbook on Integration for policy-makers and practitioners was published in April 2010 and the Commission presented an agenda for new skills and jobs entitled “A European contribution towards full employment” (COM (2010) 682 final). In 2011 a New Agenda for Integration of third country nationals will be launched, to provide improved structures and tools to facilitate the exchange of knowledge, collaboration and the mainstreaming of integration priorities of the Member States. 2.2

Harmonization in Practice – The IMP2.0 Experience

Despite the above mentioned collaboration efforts at European level, the adopted national employment migration policies are still divergent and they do not comply with the E.U. Council directives. As a result the processes and documents followed to implement common E.U. migration employment procedures (e.g. issuing of work permits) at national level are neither cross-recognized nor cross-accepted [12]. During the IMP2.0 project the five E.U. countries involved (Germany, Italy, Spain, Greece and Estonia) found that the processes followed to carry out their three common employment migration procedures (Issuing residence permits; Recognition of qualification achieved in non-E.U. countries; Estimating the number of residence permits that will be issued annually) were all sustainably different [3]. These procedures were based on particular national laws that they were not harmonized with the E.U. directives. For example, regarding the first common migration employment procedure, Issuing residence permits for dependent employment; the five IMP2.0 involved countries exert different processes for granting legal access to the labor market for third-country nationals. With respect to documents type (e.g. individual work permit, residence permit for employment), involved agents and governmental level, legal conditions, order of consecutive process-steps etc. the national processes vary considerably [3] (see e.g. in Figs. 1,2 the different processes followed in Greece and Germany).

48

N. Polemi, G. Pentafronimos, and T. Ntouskas

Greek Employer Employment contract with a remuneration at least equivalent to the monthly salary of unskilled workers.

Taxation Service Certification that they have fulfilled their tax obligations;

Municipality/Community of place of domicile/residence The third-country national submits an application and documentation showing:

Regional Agency for Aliens and Migration Examination of applications

1. That they have fulfilled their tax obligations; 2. That the requirements of employment are met;

Fig. 1. Greek Procedure [3, p.50]

Fig. 2. German Procedure [3, p.76]

Applicant

Third-country nationals who have obtained a visa for dependent employment in Greece

Residence permits for dependent employment

IMP2.0 Migration Collaborative Platform and e-Services

49

The variation of heterogeneous processes, different types and formats of documents involved result to the non-cross border recognised procedures not allowing the free-movement of legal residents (e.g. legal immigrants, second generation immigrants) in the European continent; damaging the common efforts to control the increasing migration employment wave throughout the European continent, to better balance the European labor marker and to enhance the European competiveness in a holistic way. IMP2.0 finally concluded that the increased number of decision makers and stakeholders involved in migration policies, increased and diverged legislations and implemented migration procedures, different organizational structures, large and inhomogeneous legacy systems involved, cause a “chaos” in the monitoring of national policies and the harmonization of procedures, processes and data formats. Traditional decision making instruments (e.g. face-to face meetings, workshops and conferences) at political level need to be supported with new innovative governmental IT tools (e.g. for policy modeling, business process management, visualization, collaborative decision making) in order to accelerate the harmonization efforts.

3

Societal Aspects – The IMP2.0 Requirements’ Analysis

The two migration communities (stakeholders and legal immigrants) need to be involved in the processes of drafting and adopting harmonized realistic policies. IMP2.0 used an explorative multi-methodological approach to realize their requirements and needs consisting of a workshop [5], desk research (results are outlined in the previous section), and questionnaires for both communities in various languages. In this section we will present the European level analysis of the IMP2.0 questionnaires (for the national level analysis and more details the reader is referred to [2]). IMP2.0 integrated an open-source software tool (based on the PHP framework Symphony 1.0-www.symfony-project.org-) that enables the design, development and deployment of on-line questionnaires. The IMP2.0 questionnaire tool comprises two main modules: One is responsible for the management of online questionnaires while the other undertakes the collection of answers, extraction of results and the generation of statistical data and related reports. The privacy of the responses as well as the effective management of all users’ interactions with all the systems, components and software tools, in terms of security, privacy and trust, in the IMP2.0 web platform is undertaken from an advanced, standards-based (e.g. WS-Federation, WS-Trust), centralized and federated Identity and Access Management (IAM) system, which is based on the open source implementation of OpenSSO [9]. Two types of questionnaires [1] were generated in order to capture the requirements of both migration communities and they were translated in the native languages of the above countries and also in Turkish, Albanian and Russian (the mother tongue of the majority of legal immigrants in Germany, Greece and Estonia respectively).

50

N. Polemi, G. Pentafronimos, and T. Ntouskas

Altogether four hundred and sixty eight (468) stakeholders and legal immigrants participated in the IMP2.0 questionnaire process from the five participating countries (see Table 1). Table 1. Sample of the Requirement Analysis [2, p.25] Country

Stakeholders

Immigrants

Total

Greece

22

140

162

Italy

14

111

125

Germany

17

85

102

Estonia

15

51

66

Spain

2

11

13

Total

73

398

468

The European level IMP2.0 requirements are summarized in the next subsection (for the extensive IMP2.0 requirements analysis the readers is referred to [2]). 3.1

Legal Immigrants Requirements

Information: Migrants need more and better quality information about the practices and procedures in migration/ integration/ employment field. Information has to be clear and simple (less bureaucratic), comparable (between different countries) and presented in more languages. Access to information via electronic channels, especially web, is desired. But it is important to keep in mind that in addition to electronic media, more conventional means for providing information (such as press for example) have to be used in parallel. Most needed information is about employment/ business opportunities, job/ business regulations and migration/ labor market legislation. Collaboration: Immigrants are interested in participating and collaborating among them and with the policy makers in the policy making process. The goal to include migrants in the policy formulation process could be further promoted by focusing topics of special concern such as registration and recognition of skills and qualification, restrictive regulations of work and residence permit. Migrants are interested to propose their ideas and suggestions to improve the practices and policies in the migration/ integration/ labor field, i.e. to participate in more collaborative policy-making, so possibilities for providing feedback to migration organizations have to be created. Areas for action are for example: registration and recognition of skills and qualification, restrictive regulations of work and residence permit, lack of information and communication, high bureaucracy. To increase the migrants’ participation in policy-making and improvement of procedures, room for debates and discussions should be created, where migrants/ migrant groups/ Non Governmental Organizations (NGOs) could interact with each other as well as with public administrators and policy makers exercise a two-way communication.

IMP2.0 Migration Collaborative Platform and e-Services

51

Training: Training in the language and culture of the hosting language and culture is among their main concern. The available training channels are either inadequate or very expensive and time consuming requiring high bureaucracy. Migrants are not well aware and feel confused about the possibility of processing their migration related documents online. Therefore more information, guiding and trainings need to be offered to popularize these services as well the services need to be simplified. Migrants are most interested in residence and work permit procedures. 3.2

Stakeholders Requirements

Migration policies (political level): With respect to economic/labor Migration policies stakeholders revealed a general demand for a comprehensive strategic approach for managing economic immigration. For instance, a lack of knowledge on political programmes to manage labor migration can be assessed as an indicator for the absence of a comprehensive strategic approach: Solely 25% of the stakeholders know a past, running or planned labor migration Programme in their home country. Also, it was found that stakeholders identify the need for enhancement of labor migration monitoring and means to efficiently match labor demand and supply. With respect to integration policy requirements a general demand for exchange on and further development of integration policies was identified. Practical work (practical level), Institutional cooperation requirements: Stakeholders in all participating IMP2.0 countries require internet-based structures for cooperation or dialogue with international organizations and EU-institutions, especially on labor migration. The conclusion, inter alia, is based on the finding that almost unanimously, stakeholders in all IMP2.0 participating countries ascribed high importance to international cooperation in the area of labor migration. It was found that the vast majority of stakeholders do not participate in joint projects addressing labor migration. Stakeholders require cooperation on a wide range of issues. The two most important once seem to be labor migration and exchange of good practices. Further issues of cooperation are: harmonization of policies, practices and procedures, legislation, statistical and analytical data, recognition of qualifications, migration technologies, national procedures, migration projects, illegal employment, labor market and entrepreneurship, residence and work permits, social security issues, education, problems and priorities of political intervention, integration supporting programmes, monitoring of migrant employment, monitoring of migrant selfemployment, targeted training on migration issues, simplification of documents. Process requirements: Stakeholders require communal and regional networks linking different migration and integration agents for instance, communal and regional promotion and coordination of local policies, participation of citizens, and dissemination of good practices. They also revealed that these processes would need to be enhanced by fostering networks throughout all different governmental levels also including civil society agents. Administrative procedures requirements: Stakeholders would appreciate an infrastructure for electronically processing of administrative migration related issues

52

N. Polemi, G. Pentafronimos, and T. Ntouskas

such as application or renewal of civil status documents related to employment (e.g. work permits, residence permits, family reunification documents). Information requirements referring to the information provider: A clear claim of stakeholders in three out of five countries included is not to “reinvent the wheel”. They rather require an “umbrella portal” for information sources instead of another source presenting single information. As stakeholders said, the umbrella, of course, should not solely apply for information also for other online services, such as service for matching labor demand and supply. Specific requirements: Political areas for which stakeholders see a need for action are: the development or enhancement of a comprehensive national strategic approach for managing economic immigration; the enhancement of labor migration monitoring; efficiently match labor demand and supply; exchange on and further development of policies addressing the integration of migrants into the economy. The IMP2.0 requirements’ analysis revealed the design of the IMP2.0 new emigration services described in the next section.

4

The IMP2.0 Collaborative e-Migration Services

Based on the IMP2.0 user requirements analysis described in the previous section IMP2.0 project defined a number of advanced and highly-sophisticated e-migration services [3] that aim to support both legal immigrants and migration stakeholders in their difficult efforts for effective integration and collaborative policy making and harmonization respectively. The main services to be piloted and offered by the IMP2.0 web platform [6], [7], [8] are depicted in the following figure along with the identified user requirements for each target group:

Legal immigrants

Acquire Information Evaluate Proposals Provide Feedback

Articulate Ideas / Problems / Anticipations

Information Management

Participate in the shaping of policies

Knowledge Harvesting and Content Extraction Services

Report their skills Record / Imprint their Opinions

Open Surveys Support Services

Stakeholders Post Policies and Procedures Model Policies and Procedures Analyze Policies and Procedures Harmonize Policies and Procedures

Monitor of labour migration Generate new collaboration channels Build Synergies

Build Standardized formats

Migration Policy Synchronization and Homogenization Service

Incorporate migrants in labour markets

Fig. 3. IMP2.0 User requirements and matching services

IMP2.0 Migration Collaborative Platform and e-Services

53

As depicted in the figure above, in order to satisfy the identified requirements IMP2.0 designed and implemented the following e-migration services: Information Management Services (IMS): IMP2.0 integrated a user-friendly and robust content management system serving as a vehicle for educating end-users with respect to existing policies and regulations, as well as other valuable but distributed information. Specifically, both groups of IMP2.0 end-users (i.e. legal residents, migration stakeholders) will benefit from ubiquitous and user-friendly web interfaces in order to contribute their valuable information sources (i.e. electronic documents, press reports, web links, digitalized migration-related civil status documents, etc.). Also they will be able to search and access these information assets and efficiently classify them based on a properly designed IMP2.0 taxonomy (see [4], Section 2.3.1.1). It will provide seamless access to a sustainable, scalable, structural and searchable inventory, focusing on migration related topics (e.g. National migration policies/legislation/best practices). This service provides an intuitive and simple interface that enables users to upload their contributions by providing a set of valuable and required (through the use of appropriate control mechanisms) metadata which describe each insertion based on a number of indispensable characteristics. This set of metadata includes the title and description of each information asset, a web-link when applicable as well as a number of indices that characterize the content of each contribution tagged through the integrated tree structure that represents the IMP2.0 taxonomy. The taxonomy has been defined based on the knowledge and experience in migration issues of IMP2.0 partners and has been developed using XML and javascript technologies so as to provide the desired levels of flexibility, extensibility and ease of use when accessed by the end-users for semantically tagging the content they contribute to the IMP2.0 digital library.

Fig. 4. Knowledge Harvesting and Content Extraction Services

54

N. Polemi, G. Pentafronimos, and T. Ntouskas

Knowledge Harvesting and Content Extraction Services (KHCES) (Fig. 4): will integrate advanced data collection technologies and tools as well as interactive frontend user interfaces in order to collect “Open Intelligence” and exploit the “wisdom of the crowds”. Specially, it will provide the means for collecting unstructured information and gathering different opinions, views and experiences on migration policies, national practices and procedures regarding civil status documents. This will be achieved by developing a trusted collaborative environment composed by web 2.0 tools like wikis, blogs, etc. or through graphic models thus enabling open participation, direct and multidimensional feedback, Open discussions, as well as the opportunity for gaps identification and proposal of adequate solutions. Open Surveys Support Services (OSSS): On-line questionnaires, surveys, polls, collaborative documents, and open debates will be used to compose and provide this service to both groups of end-users (i.e. legal residents, migration stakeholders). By accessing and consuming this type of services, IMP2.0 users will be able to exchange diverse proposals, support controversial ideas, positive and negative opinions, answer multiple choice questions and provide both open and structured reasoning as well as their valuable knowledge and experience. This will be achieved by integrating open-source software tools and specialized Web2.0 components which will enable distributed information acquisition and centralized data management and provide capabilities for statistical analysis and production of reports. The online IMP2.0 questionnaire tool has already been as described in the previous section. Migration Policy Synchronization and Homogenization Service (MPSHS): This service enables migration policy and decision makers, public administrators, politicians and experts to work and collaborate towards harmonized E.U. migration policies and procedures. Specifically, the objective of this service is to support migration stakeholders in posting and drafting migration policy text, developing and testing policy models, evaluation scenarios (“what-if process”) and accessing valuable and efficiently processed statistical data. Migration stakeholders (notably decision and policy makers) will be able to benchmark national practices, policies, procedures and studies as well as to perform assessments in terms of legislation, organizational, political and security characteristics, identify penetration/acceptance rates and effectively forecast, in order to overcome existing obstacles with respect to E.U. directives and best practices. To achieve these, the ImmigrationPolicy2.0 workplan takes provision for technical implementation activities, which are mainly focused on the integration of readily available R&D open source and free-license components. The main objectives of the technical implementation and integration tasks of the project are two-fold: • To establish a highly agile automation environment, that could boost both the reengineering and the integration of migration related processes. This environment will facilitate changes in the migration related processes and services, through enabling the flexible re-composition of complex policy-making and/or harmonization migration services.

IMP2.0 Migration Collaborative Platform and e-Services

55

• To compose and deploy the services and processes required to operate the project’s pilot services. The composition will leverage the SOA environment and BPM tool of the ImmigrationPolicy2.0 platform. • To provide a number of tools facilitating citizens participation, government modelling and policy development, as well as support for Open Surveys. The first step of the integration efforts of the project involves the development of a basic backbone infrastructure for the project pilots. This infrastructure comprises a portal (i.e., consisting of Web2.0 components) enabling end-users participation and interaction, a CMS enabling the management of migration information assets and related portal content, as well as the middleware infrastructure (i.e. application servers, enterprise service bus) that will provide the SOA environment. Notably, this basic infrastructure will hold the data structures of the project including taxonomies, the XML-based and e-Gif enabled content objects, while also providing related tools for the effective collection of data and simple query execution (e.g. online forms, questionnaires etc.) as well. In addition it will provide: • Advanced content management tools such as rich text editors, live page editing and scheduling, and advanced document managers. • Web 2.0 aware technologies with their own set of authorizations, message boards for facilitating conversations around migration-based topics, blogs for allowing users to convey information and RSS feeds from the last mentioned message boards and blogs within ImmigrationPolicy2.0. • A multi-tier search engine so that end-users are able to search relevant information throughout an entire web interactive portal, within specific portlets such as Wikis, Message Boards, other Web 2.0 aware technologies and even in external integrated applications through its advanced multiple interfacing module. • Intuitive front end user interfaces that will share a set of common characteristics to promote user friendliness and accessibility: they will be multilingual, in order for users to easily toggle between different language settings and they will follow standardized best practices for accessibility (especially for special groups). • Web publishing tools so to easily create and manage content, from a simple article of text and images to fully functional web sources. Summarizing, IMP2.0 implements a range of highly sophisticated services in order to provide a robust and sustainable IS/IT collaboration framework that is able to support decision-making in the critical domain of migration. Integrating advanced tools for collecting, analyzing and sharing diverse information and supporting policy-making decisions by providing the opportunity for all key players to work together and contribute their knowledge and experience, the proposed solution delivers an efficient synchronization framework capable to leverage existing traditional practices.

5

Conclusions

For more than 20 years now (from the Maastricht Treaty, 1986, to the Treaty of Amsterdam in force since 1999) the Member States have joined forces to combat

56

N. Polemi, G. Pentafronimos, and T. Ntouskas

international phenomena such as illegal immigration. There is an urgent, (acknowledged by the ongoing Amsterdam Treaty and the E.U. Councils’ programmes) need for more effective, collaborative actions among Member States in order to harmonize national migration policies and specific migration procedures. The technological, policy-driven, IMP2.0 project contributes directly to these needs and responds to the invitation of the E.U. Council for the development of information systems that will contribute towards a common E.U. migration policy. IMP2.0 performed a social analysis on the migration needs and requirements involving both stakeholders and legal immigrants in order to propose e-migration services considering all aspects (policy, legal, societal). Collaborative technologies are the IMP2.0 technological tools used in order to offer e-migration collaborative services hosted in its Web2.0 based-platform. The IMP2.0 user requirement analysis revealed the need for various new e-migration services. The IMP2.0 serves as an innovative migration system able to host additional emigration services for legal immigrants (e.g. training/ consulting, employment opportunities for legal immigrants, on line applications for labor migration documents, recognition of qualifications, certifications) and stakeholders (e.g. matching employment supply-demand, labor migration monitoring, online processing of migration related documents, one-stop entrance on migration information ) serving their needs and expanding the e-society concept to all its participants. Acknowledgments. The authors are grateful to the E.C. ICT PSP Programme for funding the ImmigrationPolicy2.0 (IMP2.0) project and all the partners for their valuable contribution. Especial thanks to Professor Wolfgang Bosswick and Verena Krobisch from EFMS for their substantial input in the societal analysis.

References 1. European Project Immigration Policy 2.0, http://www.immigrationpolicy2.eu 2. Bosswick, W., Krobisch, V., Ntouskas, T., Pentafronimos, G., Polemi, N., et al.: D1.1 End User Requirements. European project deliverable, Immigration Policy 2.0 (2011), http://www.immigrationpolicy2.eu/?q=node/22 3. Pentafronimos, G., Ntouskas, T., Polemi, N., Rodríguez, N., Bosswick, W., et al.: D2.1 User and Service Classification. European project deliverable, Immigration Policy 2.0 (2011), http://www.immigrationpolicy2.eu/?q=node/22 4. Rodríguez, N., Ntouskas, T., Pentafronimos, G., Polemi, N., Krobisch, V., et al.: D1.2 State-of-the-art and advances in European migration policies. European project deliverable, Immigration Policy 2.0 (2010), http://www.immigrationpolicy2.eu/?q=node/22 5. First Immigration Policy 2.0 Workshop, http://www.immigrationpolicy2.eu/?q=node/21 6. Pentafronimos, G., Karantjias, A., Polemi, N.: ODYSSEUS: An Advanced, Collaborative and Trusted Framework for the Provision of Migration Services. In: Fifth International Conference on the Internet and Web Applications and Services (ICIW), pp. 531–537. IEEE Computer Society, Barcelona (2010)

IMP2.0 Migration Collaborative Platform and e-Services

57

7. Pentafronimos, G., Karantjias, A., Polemi, N.: OPSIS: An Open, Preventive and Scalable Migration Information System. In: Third International Conference on e-Democracy (eDemocracy), pp. 341–350. Springer Lecture Notes, Athens (2009) 8. Karantjias, A., Polemi, N., Pentafronimos, G.: Participatory Immigration Policy Making and Harmonization Services based on Collaborative Web 2.0 Technologies. In: Ahson, S., Ilyas, M. (eds.) Service Delivery Platforms: Developing Converged Multimedia Services HandBook. CRC Press, Taylor & Francis Group Publishing (March 2011) 9. OpenSSO, http://java.net/projects/opensso 10. Kate, M.-A., Niessen, J.: Guide to Locating Migration Policies in the European Commission. EPIM and MPG, ed. (2008), http://www.epim.info/docs/documents/137.GuidetoLocatingImmig rationPoliciesintheECII_31.10.08.pdf 11. Collet, E.: The European Union’s Stockholm Program: Less Ambition on Immigration and Asylum, But More Detailed Plans. Migration Policy Institute, Migration Information Source (2010), http://www.migrationinformation.org/Feature/display.cfm?ID=768 12. Fleischmann, F., Dronkers, J.: The effects of social and labour-market policies of EUcountries on the socio-economic integration of first and second generation immigrants from different countries of origin. In: EUI Working Paper, RCSAS 2007/19, European University Institiute, Florence, Robert Schuman Centre for Advanced Studies (2007) 13. Council Regulation EC 109/2003: Concerning the status of third-country nationals who are long-term residents. Official Journal of the European Union L 16, 44–53 (January 23, 2004) 14. Council Directive EC 86/2003: On the right to family reunification. Official Journal of the European Union L 251, 12–18 (October 03, 2003)

Consumer Side Resource Accounting in the Cloud Ahmed Mihoob, Carlos Molina–Jimenez, and Santosh Shrivastava School of Computing Science, Newcastle University Newcastle upon Tyne, NE1 7RU, UK {a.m.mihoob,carlos.molina,santosh.shrivastava}@ncl.ac.uk

Abstract. The paper investigates the possibility of consumer side resource accounting of cloud services, meaning, whether it is possible for a consumer to independently collect all the resource usage data required for calculating billing charges for pay-per-use cloud services. The accounting models of two widely used cloud services are examined and possible sources of difficulties are identified, including causes that could lead to discrepancies between the metering data collected by the consumer and the provider. The investigation is motivated by the fact that cloud service providers perform their own measurements to collect usage data, but as yet there are no equivalent facilities of consumer-trusted metering that are commonly provided by utility service providers like gas and electricity. The paper goes on to suggest how cloud service providers can improve their accounting models to facilitate the task of consumer side resource accounting. Keywords: cloud resource consumption, storage and computational resources, resource metering and accounting models, Amazon Web Services.

1

Introduction

Cloud computing services made available to consumers range from providing basic computational resources such as storage and compute power (infrastructure as a service, IaaS) to sophisticate enterprise application services (software as a service SaaS) A common business model is to charge consumers on a pay-peruse basis where they periodically pay for the resources they have consumed. Needless to say that for each pay-per-use service, consumers should be provided with an unambiguous resource accounting model that precisely describes all the constituent chargeable resources of the service and how billing charges are calculated from the resource usage (resource consumption) data collected on behalf of the consumer over a given period. If the consumers have access to such resource usage data then they can use it in many interesting ways, such as, making their applications billing aware, IT budget planning, create brokering services that automate the selection of services in line with user’s needs and so forth. Indeed, it is in the interest of the service providers to make resource consumption data available to consumers; incidentally all the providers that we know of do make such data accessible to their consumers in a timely fashion. T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 58–72, 2011. c IFIP International Federation for Information Processing 2011 

Consumer Side Resource Accounting in the Cloud

59

An important issue that is raised is the accountability of the resource usage data: who performs the measurement to collect the resource usage data the provider, the consumer, a trusted third party (TTP), or some combination of them1 ? The traditional utility providers such as water, gas and electricity perform their own measurements to collect usage data using metering devices (trusted by consumers) that are deployed in the consumers premises. Cloud service providers also perform their own measurements to collect usage data, although, as yet there are no equivalent facilities of consumer-trusted metering; rather, consumers have no choice but to take whatever usage data that is made available by the provider as trustworthy. A good introduction into the underlying trust issues can be found in [12]. In light of the above discussion, it is worth investigating whether it is possible for a consumer (or a TTP acting on behalf of the consumer) to independently collect all the resource usage data required for calculating billing charges. In effect, this means that a consumer (or a TTP) should be in a position to run their own metering service for measuring resource consumption. If this were possible, then consumers will be able to perform reasonableness checks on the resource usage data available from service providers as well as raise alarms when apparent discrepancies are suspected in consumption figures; furthermore, innovative charging schemes can be constructed with confidence by consumers who are themselves offering third party brokering services. In our earlier paper [11], we developed the notion of a Consumer–centric Resource Accounting Model for a cloud service. We say that a resource accounting model is weakly consumer– centric if all the data that the model requires for calculating billing charges can be queried programmatically from the provider. Further, we say that an accounting model is strongly consumer–centric if all the data that the model requires for calculating billing charges can be collected independently by the consumer (or a TTP). Strongly consumer–centric accounting models have the desirable property of openness and transparency, since service users are in a position to verify the charges billed to them. That paper also evaluated the accounting model of Simple Storage Service, S3 from Amazon to see how well it matches the proposed notion. This paper contributes to the prior work in three ways: (i) the evaluation work on accounting models is extended to include a compute service (Amazon Elastic Compute Cloud, EC2) and we point out a few ambiguities in the EC2 model description (Section 3); (ii) we precisely identify the causes that could lead to discrepancies between the metering data collected by the provider and the consumer, and whether the discrepancies can be resolved (Section 4); and (iii) we present ideas on how an accounting model should be constructed so as to make them strongly consumer–centric (Section 5).

1

A note on terminology: ’accountability’ refers to concepts such as responsibility, answerability, trustworthiness; not to be confused with ’resource accounting’ that refers to the process concerned with calculating financial charges.

60

2

A. Mihoob, C. Molina–Jimenez, and S. Shrivastava

Background

For resource accounting it is necessary to determine the amount of resources consumed by a given consumer (also called client and subscriber) during given time interval, for example, a billing period. Accounting systems are composed of three basic services: metering, accounting and billing. provider service interface

consumer

. application

metering metering data service

accounting service accounting model

accounting data

billing service

billing data

accounting system

Fig. 1. Consumer side resource accounting system

We show a consumer side accounting system in Fig.1. We assume that resources are exposed as services through one or more service interfaces. As shown in the figure, the metering service intercepts the message traffic between the consumer application and the cloud services and extracts relevant data required for calculating resource usage (for example, the message size which would be required for calculating bandwidth usage). The metering service stores the collected data for use by the accounting service. The accounting service retrieves the metering data, computes resource consumption from the data using its accounting model and generates accounting data that is needed by the billing service to calculate the billing data. Accounting models are provider–specific in the sense that the functionality of an accounting model is determined by the provider’s policies. These policies determine how the metrics produced by his metering service are to be interpreted; for example, 1.7 GB of storage consumption can be interpreted by the provider’s accounting model either as 1 or 2 GB. The accounting models of cloud providers are normally available from their web pages and in principle can be used by subscriber to perform their own resource accounting. The difficulty here for the subscriber is to extract the accounting model from their online documentation as most providers that we know of, unnecessarily blur their accounting models with metering and billing parameters. The parameters involved in accounting models depend on the type of service (SaaS, PaaS, IaaS, etc.) offered. In this paper we will examine, from the point of view of consumer side resource accounting, the accounting models of Amazons Simple Storage Service (S3) and Elastic Compute Cloud (EC2). In the following discussion, we gloss over the fine details of pricing, but concentrate on metering and accounting services.

Consumer Side Resource Accounting in the Cloud

3 3.1

61

Accounting of Resource Consumption S3 Accounting Model

An S3 space is organised as a collection of buckets which are similar to folders. A bucket can contain zero or more objects of up to 5 terabytes of data each. Both buckets and objects are identified by names (keys in Amazon terminology) chosen by the customer. S3 provides SOAP and RESTful interfaces. An S3 customer is charged for: a) storage: storage space consumed by the objects that they store in S3; b) bandwidth: network traffic generated by the operations that the customer executes against the S3 interface; and c) operations: number of operations that the customer executes against the S3 interface. Storage: The key parameter in calculation of the storage bill is number of byte hours accounted to the customer. Byte Hours (ByteHrs) is the the number of bytes that a customer stores in their account for a given number of hours. Amazon explains that the GB of storage billed in a month is the average storage used throughout the month. This includes all object data and metadata stored in buckets that you created under your account. We measure your usage in TimedStorage–ByteHrs, which are added up at the end of the month to generate your monthly charges. They further state that at least twice a day, we check to see how much storage is used by all your Amazon S3 buckets. The result is multiplied by the amount of time passed since the last checkpoint. From the definition of ByteHrs it follows that to calculate their bill, a customer needs to understand 1) how their byte consumption is measured, that is, how the data and metadata that is uploaded is mapped into consumed bytes in S3; and 2) how Amazon determines the number of hours that a given piece of data was stored in S3 —this issue is directly related to the notion of a checkpoint. Amazon explains that each object in S3 has, in addition to its data, system metadata and user metadata; furthermore it explains that the system metadata is generated and used by S3, whereas user metadata is defined and used only by the user and limited to 2 KB of size [1]. Unfortunately, Amazon does not explain how to calculate the actual storage space taken by data and metadata. To clarify the issue, we conducted a number of experiments (see [11]) involving uploading of a number of objects of different names, data and user metadata into an equal number of empty buckets. Three conclusions can be drawn from these experiments: first, the mapping between bytes uploaded (as measured by intercepting upload requests) and bytes stored in S3 correspond one to one; second, the storage space occupied by system metadata is the sum of the lengths (in Bytes) of object and bucket names and incur storage consumption; third, user metadata does not impact storage consumption. In summary, for a given uploaded object, the consumer can accurately measure the total number of bytes that will be used for calculating ByteHrs. Next, we need to measure the ’Hrs’ of ’ByteHrs’. As stated earlier, Amazon states that at least twice a day they check the amount of storage consumed by a customer. However, Amazon does not stipulate exactly when the checkpoints take place.

62

A. Mihoob, C. Molina–Jimenez, and S. Shrivastava

To clarify the situation, we conducted a number of experiments that consisted in uploading to and deleting files from S3 and studying the Usage Reports of our account to detect when the impact of the PUT and DELETE operations were accounted by Amazon. Our findings are summarised in Fig.2. It seems that, currently, Amazon does not actually check customers’ storage consumption twice a day as they specify in their Calculating Your Bill document, but only once. From our observations, it emerged that the time of the checkpoint is decided randomly by Amazon within the 00:00:00Z and 23:59:59Z time interval2 . In the figure, CP stands for checkpoint, thus CP30 : 2GB indicate that CP30 was conducted on the 30th day of the month at the time specified by the arrow and reported that at that time the customer had 2 GB stored in S3. SC stands for Storage Consumption and is explained below.

30

31 Mar

01

CP : 2GB

CP : 5GB

SC for the 30th 2x24=48GBHrs

SC for the 31st SC for the 1st 5x24=120GBHrs 3x24=72GBHrs

30

31

02

CP : 3GB 01

Fig. 2. Amazon’s checkpoints

As shown in the figure, Amazon uses the results produced by a checkpoint of a given day, to account the customer for the 24 hrs of that day, regardless of the operations that the customer might perform during the time left between the checkpoint and the 23:59:59Z hours of the day. For example, the storage consumption for the 30th will be taken as 2× 24 = 48 GBHrs; where 2 represents the 2GB that the customer uploaded on the 30th and 24 represents the 24 hrs of the day. Bandwidth: Amazon charges for the network data transferred from the customer to S3 (’DataTransfer–In’, the request message of an operation) and the network data transferred from S3 to the customer (the corresponding response message, ’DataTransfer–Out’). It is however not clear from the available information how the size of of the message is calculated. We therefore conducted several experiments involving uploading, downloading, deleting etc. of objects using both RESTful and SOAP interfaces and compared the information extracted from the intercepted messages with the information available from Amazon usage reports. It turns out that for RESTful operations, only the size of the object (in DataTransfer–In for PUT, and DataTransfer–Out for GET) is taken into account and system and user metadata is not part of the overhead, whereas for SOAP operations, the total size of the message is taken into account. 2

S3 servers are synchronised to the Universal Time Coordinated (UTC) which is also known as the Zulu Time (Z time) and in practice equivalent to the Greenwich Mean Time (GMT).

Consumer Side Resource Accounting in the Cloud

63

Operations: It is straightforward for a consumer to count the type and number of operations performed on S3. We note that an operation might fail to complete successfully. The error response in general contains information that helps identify the party responsible for the failure: the customer or the S3 infrastructure. For example, NoSuckBucket errors are caused by the customer when they try to upload a file into a non-existent bucket; whereas an InternalError code indicates that S3 is experiencing internal problems. Our understanding is that the consumer is charged for an operation, whether the operation succeeded or not. 3.2

EC2 Accounting Model

EC2 is a computation service offered by Amazon as an IaaS [4]. The service offers raw virtual CPUs (also called a Virtual Machine or VM) to subscribers. A subscriber is granted administrative privileges over his VM, that he can exercice by means of sending remote comands to the Amazon Cloud from his desktop computer. For example, he is expected to configure, launch, stop, re–launch, terminate, backup, etc. his VM. In return, the subscriber is free to choose the operating system (eg Windows or Linux) and applications to run. In EC2 terminology, a running virtual CPU is called an instance whereas the frozen bundle of software on disk that contains the libraries, applications and initial configuration settings that are used to launch an instance is called the Amazon Machine Image (AMI). Currently, Amazon offers six types of instances that differ from each other in four initial configuration parameters that cannot be changed at running time: amount of EC2 compute units that it delivers, size of their memory and local storage (also called ephemeral and instance storage) and the type of platform (32 or 64 bits). An EC2 compute unit is an Amazon unit and is defined as the equivalent CPU capacity of a 1.0–1.2 GHz 2007 Opteron or 2007 Xeon processor. Thus Amazon offer small, large, extra large and other types of instances. For example, the default instance type is the Small Instance and is a 32 bit platform that delivers 1 EC2 compute unit and provided with 1.7 GB of memory and 160 GB of local storage. These types of instances are offered to subscribers under several billing models: on–demand instances, reserved instances and spot instances. In our discussion we will focus on on–demand instances. Under the on–demand billing model, Amazon defines the unit of consumption of an instance as the instance hour (instanceHrs). Currently, the cost of an instance hour of a small instance running Linux or Windows, is, respectively, 9.5 and 12 cents. On top of charges for instance hours, instance subscribers normally incur additional charges for data tranfer that the instances generates (Data Transfer In and Data Transfer Out) and for addtional infrastructure that the instance might need such as disk storage, IP addresses, monitoring facilities and others. As these additional charges are accounted and billed separately, we will leave them out of our discussion and focus only on instance hours charges.

64

A. Mihoob, C. Molina–Jimenez, and S. Shrivastava

The figures above imply that if a subscriber accrues 10 instanceHrs of a small instance consumption, running Linux, during a month, he will incur a charge of 95 cents at the end of the month. In principle, the pricing tables publicly available from Amazon web pages should allow a subscriber to independently conduct his own accounting of EC2 consumption. In the absence of a well defined accounting model this is not a trivial exercise. Insights into the EC2 accounting model are spread over several on–line documents from Amazon. Some insight into the definition of instance hour is provided in the Amazon EC2 Pricing document [3] (see just below the table of On–demand Instances) where it is stated that Pricing is per instance–hour consumed for each instance, from the time an instance is launched until it is terminated. Each partial instance–hour consumed will be billed as a full hour. This statement suggests that once an instance is launched it will incur at least an instance hours of consumption. For example, if the instance runs continuouly for 5 minutes, it will incur 1 instanceHrs; likewise, if the instance runs continuously for 90 minutes, it will incur 2 instanceHrs. The problem with this definition is that it does not clarify when an instance is considered to be launched and terminated. Additional information about this issue is provided in the Billing section of FAQs [2], Paying for What You Use of the Amazon Elastic Compute (Amazon EC2) document [4] and in the How You’re Charged section of the User Guide [5]. For example, in [4] it is stated that Each instance will store its actual launch time. Thereafter, each instance will charge for its hours of execution at the beginning of each hour relative to the time it launched. From information extracted from the documents cited above it is clear that Amazon starts and stops counting instance hours as the instance is driven by the subscriber, through different states. Also, it is clear that Amazon instance hours are accrued from the execution of one or more individual sessions executed by the subscriber during the billing period. Within this context, a session starts and terminates when the subscriber launches and terminates, respectively, an instance. Session–based accounting models for resources that involve several events and states that incur different consumptions, are conveniently described by Finite State Machines (FSMs). We will use a FSM to describe EC2 accounting model. States of an instance session. The states that an instance can reach during a session depend on the type of memory used by the AMI to store its boot (also called root) device. Currently, Amazon supports S3–backed and EBS–backed instances. EBS stands for Elastic Block Store and is a persistent storage that can be attached to an instance. The subscriber chooses between S3 or EBS backed instances at AMI creation time. Unfortunatelly, the states that an instance can reach during a session are not well documented by Amazon. Yet after a careful examination of Amazon’s online documentation we managed to build the FSM shown in Fig. 3–a).

Consumer Side Resource Accounting in the Cloud

AMI configured

AMI configured launch/ NinsHrs=1; timer=0; starttimer launch/ NinsHrs=1; timer=0; starttimer pending

launch/

reboot

failure/ timer=0

failure/ timer=0

NinsHrs=1; timer=0; starttimer

application commands stop/ timer=0

running

terminate/ timer=0

stopping

stopped

rebooting

timer=60min/ NinsHrs++; timer=0; starttimer

a)

reboot

failure/ timer=0

shutting down

terminated

terminate

launch/

pending

application commands rebooting

65

failure/ timer=0

stop/ timer=0

running

terminate/ timer=0

stopping

stopped

timer=60min/ NinsHrs++; timer=0; starttimer

shutting down

terminated

terminate

b)

Fig. 3. Session of an Amazon instance represented as FSM

The FSM of an Amazon instance includes two types of states: permanent and transient states. Permanent states (represented by large circles, e.g. running) can be remotely manipulated by commands issued by the subscriber; once the FSM reaches a permanent state, it remains there until the subscribers issues a command to force the FSM to progress to another state. Transient states (represented by small circles, e.g. stopping) are states that the FSM visits temporarily as it progresses from a permanent state into another. The subscriber has no control over the time spent in a transient state; this is why there are no labels on the outgoing arrows of these states. We have labeled the transitions of the FSM with event/action notations. The event is the cause of the transition whereas the action represents the set (possibly empty) of operations that Amazon executes when the event occurs, to count the numbers of instance hours consumed by the instance. There are two types of events: subscriber’s and internal to the FSM events. The subscriber’s events are the comands (launch, application commands, reboot, stop and terminate) that the subscribers issues to operate his instance; likewise, internal events are events that occur independently from the subscriber’s commands, namely, timer = 60min and failure. AMI configured: is the initial state. It is reached when the subscriber sucessfully configures his AMI so that it is ready to be launched. running: is the state where the instance can perform useful computation for the subscriber, for example, it can respond to application commands issued by the subscriber. terminated: is the final state and represents the end of the life cycle of the instance. Once this state is reached the instance is destroyed. To perform additional computation after entering this state the subscriber needs to configure another AMI. The terminated state is reached when the subscribed issues the terminate command, the instance fails when it is in running state or the instance

66

A. Mihoob, C. Molina–Jimenez, and S. Shrivastava

fails to reach running state. shuttingdown: is reached when the subscriber issues the terminate command. stopped: this state is supported only EBS–backed instances (S3–backed instances cannot be stopped) and is reached when the user issues stop command, say for example, to perform backup duties. rebooting: is reached when the subscriber issues the reboot command. States and and instance hours: In the figure, N instHrs is used to count the number of instance hours consumed by an instance during a single session. The number of instance hours consumed by an instance is determined by the integer value stored in N instHrs when the instance reaches the terminated state. timer is Amazon’s timer to count 60 minutes intervals; it can be set to zero (timer = 0) and started (starttimer). In the FSM, the charging operations are executed as suggested by the Amazon’s on line documentation. For example, in Paying for What You Use Section of [4], Amazon states that the beginning of an instance hour is relative to the launch time. Consequently, the FSM sets N instHrs = 1 when the subscriber executes a launch command from the AMI configured state. At the same time, timer is set to zero and started. N instHrs = 1 indicates that once a subscriber executes a launch command, he will incur at least one instance hour. If the subscriber leaves his instance in the running state for 60 minutes (timer = 60min) the FSM increments N instHrs by one, sets the timer to zero and starts it again. From running state the timer is set to zero when the subscriber decides to terminate his instance (terminate command) or when the instance fails (failure event). Although Amazon’s documentation does not discuss it, we believe that the possibility of an instance not reaching the running state cannot be ignore, therefore we have included a transition from pending to terminated state; the FSM sets the timer to zero when this abnormal event occurs. As explained in Basics of Amazon ESB–Backed AMIs and Instances and How You’re Charged of [5], a running ESB–backed instance can be stopped by the subscriber by means of the stop command and drive it to the stopped state. As indicated by timer = 0 operation executed when the subscribed issues a stop command, an instance in stopped state incurs no instance hours. However, though it is not shown in the figure as this is a different issue, Amazon charges for EBS storage and other addtional services related to the stopped instance. The subscriber can drive an instance from the stopped to the terminated state. Alternatively he can re–launch his instance. In fact, the subscriber can launch, stop and launch his instance as many times as he needs to. However, as indicated by the N instHrs + + , timer = 0 and starttimer operations over the arrow, every transition from stopped to pending state accrues an instance hours of consumption, irrespectively of the time elapsed between each pair of consecutive launch commands. Experiments with Amazon instances: To verify that the accounting model described by the FSM of Fig. 3–a) matches Amazon’s description, we (as subscribers) conducted a series of practical experiments. In particular, our aim was to verify how the number of instance hours is counted by Amazon.

Consumer Side Resource Accounting in the Cloud

67

The experiments involved 1) configuration of different AMIs; 2) launch of instances; 3) execution of remote commands to drive the instances through the different states shown in the FSM. For example, we configured AMIs, launched and run them for periods of different lengths and terminated them. Likewise, we launched instances and terminated them as soon as they reached the running state. To calculate the number of instance hours consumed by the instances, we recorded the time of execution of the remote commands launch, stop, terminate and reboot, and the time of reaching both transient and permanent states. For comparison, we collected data (start and end time of an instance hour, and number of instance hours consumed) from Amazon EC2 usage report. A comparison of data collected from our experiments against Amazon’s data from their usage report reveals that currently, the beginning of an instance hour is not the execution time of the subscriber’s launch command, as documented by Amazon, but the time when the instance reaches the running state. These findings imply that the current accounting model currently in use is the one described by the FSM of Fig. 3–b). As shown in the figure, the N instHrs is incremented when the instance reaches the running state.

4 4.1

Potential Causes of Discrepancies Storage

Since, for the calculation of ByteHrs, the time of the checkpoint is decided randomly by Amazon within the 00:00:00Z and 23:59:59Z time interval, the time used at the consumer’s side need not match that at the provider’s side: a potential cause for discrepancy. This is illustrated with the help of Fig.4. 30 PUT 2GB PUT 1GB PUT 3GB DEL3GB

31 Mar PUT 4GB

CP : 6GB cp : 6GB 30

01 PUT 5GB

cp : 7GB 31

30

CP : 7GB 31

30 PUT 2GB PUT 1GB PUT 3GB DEL3GB CP : 6GB 30

cp : 3GB 30

31 Mar

01

PUT 4GB cp : 3GB 31

PUT 5GB

CP : 7GB 31

SC for Mar= 6x24 + 7x24= 312GBHrs sc for Mar= 6x24 + 7x24= 312GBHrs

SC for Mar= 6x24 + 7x24= 312GBHrs sc for Mar= 3x24 + 3x24= 144GBHrs

a)

b)

Fig. 4. Impact of checkpoints

The figure shows the execution time of four PUT and one DEL operations executed by an S3 consumer during the last two days of March. The first day of April is also shown for completeness. For simplicity, the figure assumes that the earliest PUT operation is the very first executed by the consumer after opening his S3 account. The figure also shows the specific points in time when checkpoints are conducted independently by two parties, namely, Amazon and a consumer. Thus, CP and cp represent, respectively, Amazon’s and the consumer’s checkpoints; the Giga Bytes shown next to CP and cp indicate the storage consumption detected by the checkpoint. For example, on the 30th, Amazon conducted

68

A. Mihoob, C. Molina–Jimenez, and S. Shrivastava

its checkpoint about five in the morning and detected that, at that time, the customer had 6 GB stored (CP30 : 6GB). On the same day, the consumer conducted his checkpoint just after midday and detected that, at that time, he had 6 GB stored (cp30 : 6GB). SC and sc represent, respectively, the storage consumption for the month of March, calculated by Amazon and consumer, based on their checkpoints. The figure demonstrates that the storage consumption calculated by Amazon and consumer might differ significantly depending on the number and nature of the operations conducted within the time interval determined by the two parties’ checkpoints, for example, within CP31 and cp31 . Scenario a) shows an ideal situation where no consumer’s operations are executed within the pair of checkpoints conducted on the 30th or 31st. The result is that both parties calculate equal storage consumptions. In contrast, b) shows a worse–case scenario where the DEL operation is missed by CP30 and counted by cp30 and the PUT operation is missed by cp31 and counted by CP31 ; the result of this is that Amazon and the consumer, calculate SC and sc, respectively, as 312 GB and 144 GB. Ideally, Amazon’s checkpoint times should be made known to consumers to prevent any such errors. Providing this information for upcoming checkpoints is perhaps not a sensible option for a storage provider, as the information could be ’misused’ by a consumer by placing deletes and puts around the checkpoints in a manner that artificially reduces the consumption figures. An alternative would be to make the times of past checkpoints available (e.g., by releasing them the next day). Impact of network and operation latencies: In the previous discussion concerning calculation of ByteHrs (illustrated using Fig. 4), we have implicitly assumed that the execution of a PUT (respectively a DELETE) operation is an atomic event whose time of occurrence is either less or greater than the checkpoint time (i.e., the operation happens either before or after the checkpoint). This allowed us to say that if the checkpoint time used at the provider is known to the consumer, then the consumer can match the ByteHrs figures of the provider. However, this assumption is over simplifying the distributed nature of the PUT (respectively a DELETE) operation. In Fig.5 we explicitly show network and operation execution latencies for a given operation, say PUT; also, i, j, k and l are provider side checkpoint times used for illustration. Assume that at the provider side, only the completed operations are taken into account for the calculation of ByteHrs; so a checkpoint taken at time i or j will not include the PUT operation (PUT has not yet completed), whereas a checkpoint taken a time k or l will. What happens at the consumer side will depend on which event (sending of the request or reception of the response) is taken to represent the occurrence of PUT. If the timestamp of the request message (PUT) is regarded as the time of occurrence of PUT, then the consumer side ByteHrs calculation for a checkpoint at time i or j will include the PUT operation, a discrepancy since the provider did not! On the other hand, if the timestamp of the response message is regarded as the time of occurrence of PUT, then a checkpoint at time k will not include the PUT operation (whereas the provider has), again

Consumer Side Resource Accounting in the Cloud

69

a discrepancy. In short, for the operations that occur ’sufficiently close’ to the checkpoint time, there is no guarantee that they get ordered identically at both the sides with respect to the checkpoint time. i

j

k

l

Provider

Consumer

Fig. 5. Network and operation latencies

Operations: Earlier we stated that it is straightforward for a consumer to count the type and number of operations performed on S3. There is a potential for discrepancy caused by network latency: operations that are invoked ’sufficiently close’ to the end of an accounting period (say i) and counted by the consumer for that period, might get counted as performed in the next period (say j) by the provider if due to the latency, these invocation messages arrive in period j. This will lead to the accumulated charges for the two period not being the same. This is actually not an issue, as the Amazon uses the timestamp of the invocation message for resolution, so the consumer can match the provider’s figure. One likely source of difficulty about the charges for operations is determining the liable party for failed operations. Currently, this decision is taken unilaterally by Amazon. In this regard, we anticipate two potential sources of conflicts: DNS and propagation delays. As explained by Amazon, some requests might fail and produce a Temporary Redirect (HTTP code 307 error) due to temporary routing errors which are caused by the use of alternative DNS names and request redirection techniques [13]. Amazon’s advice is to design applications that can handle redirect errors, for example, by resending a request after receiving a 307 code(see [1], Request Routing section). Strictly speaking these errors are not caused by the customer as the 307 code suggests. It is not clear to us who bears the cost of the re–tried operations. 4.2

EC2

The mismatch between Amazon’s documented accounting model and the one currently in use (Fig. 3–a and b, respectively) might result in discrepances between the subscriber’s and Amazon’s calculations of instance hours. For example, imagine that it takes five minutes to reach the running state. Now imagine that the subscriber launches an instance, leaves it running for 57 minutes and then terminates it. The subscriber’s N instHours will be equal to two: N instHrs = 1 at launch time and then N instHrs is incremented when timer = 60min. In contrast, to the subscriber’s satisfaction, Amazon’s usage records will show only one instance hour of consumption. One can argue that this discrepancy is not of the subscriber’s concern since, economically, it always favours him.

70

A. Mihoob, C. Molina–Jimenez, and S. Shrivastava

More challenging and closer to the subscriber’s concern are discrepancies caused by failures. Amazon’s documentation does not stipulates how instances that fail accrue instance hours. For example, examine Fig. 3–a) and imagine that an instance suddenly crashes after spending 2 hrs and 15 min in running state. It is not clear to us whether Amazon will charge for the last 15 min of the execution as a whole instance hour. As a second example, imagine that after being launched either from AMI configured or stopped states, an instance progresses to pending state and from there, due to a failure, to terminated. It is not clear to us if Amazon will charge for the last instance hour counted by N instHrs. We believe that, apart from these omissions about failure situations, the accounting model of Fig. 3–a) can be implemented and used by the subscriber to produce accurate accounting. A salient feature of this model is that all the events (launch, stop and terminate) that impact the N instHrs counter are generated by subscriber. The only exeception if the timer = 60min event, but that can be visible to the subscriber if he synchronises his clock to UTC time. The accounting model that Amazon actually uses (Fig. 3–b) is not impacted by failures of instances to reach running state because in this model, N insHrs is incremented when the instance reaches running state. However, this model is harder for the subscriber to implement since the event that causes the instance to progress from pending to running state is not under the subscriber’s control.

5

Related Work

An architecture for accounting and billing for resources consumed in a federated Grid infrastructure is suggested in [9]. The paper provides a valuable insight into the requirements (resource re–deployment, SLA awareness, pre–paid and post– paid billing, standardised records and others) that accounting and billing services should meet. In [6], the author discuss similar requirements for accounting and billing services, but within the context of federated network of telecommunication providers. Both papers overlook the need to provide consumers with means of performing consumer–side accounting. A detailed discussion of an accounting system similar to the one shown in Fig. 1 but aimed at telecommunication services is provided in [10]. Accounting models are fundamental to subscribers interested in planning for minimisation of expenditures on cloud resources. The questions raised are what workload to outsource, to which provider, what resources to rent, when, and so on. Examples of research results in this direction are reported in [14,7]. In [7] the authors discuss how an accouting service deployed within an organisation can be used to control expenditures on public cloud resources; their accounting service relies on data downloaded from the cloud provider instead of calculating it locally. In [8], the authors take Amazon cloud as an example of cloud provider and estimate the performance and monetary–cost to compute a data–intensive (terabytes) workflow that requires hours of CPU time. The study is analytical (as

Consumer Side Resource Accounting in the Cloud

71

opposite to experimental) and based on the authors’ accouting model. For instance, to produce actual CPU–hours, they ignore the granularity of Amazon instance hours and assume CPU seconds of computation. This work stresses the relevance of accounting models.

6

Concluding Remarks

We investigated whether it is possible for a consumer (or a TTP acting on behalf of the consumer) to independently collect, for a given cloud service, all the resource usage data required for calculating billing charges. We examined two main IaaS services: storage and compute from Amazon; our investigation revealed the causes that could lead to discrepancies between the metering data collected by the the consumer not matching that of the provider. Essentially these causes can be classed into three categories discussed below. 1. Incompleteness and ambiguities: We pointed out several cases where an accounting model specification was ambiguous or not complete. For example, regarding bandwidth consumption, it is not clear from the available information what constitutes the size of of a message. It is only through experiments we worked out that for RESTful operations, only the size of the object is taken into account and system and user metadata is not part of the message size, whereas for SOAP operations, the total size of the message is taken into account. Failure handling is another area where there is lack of information and/or clarity: for example, concerning EC2, it is not clear how instances that fail accrue instance hours. 2. Unobservable events: If an accounting model uses one or more events that impact resource consumption, but these events are not observable to (or their occurrence cannot be deduced accurately by) the consumer, then the data collected at the consumer side could differ from the that of the provider. Calculation of storage consumption in S3 (ByteHrs) is a good example: here, the checkpoint event is not observable. 3. Differences in the measurement process: Difference can arise if the two sides use different techniques for data collection. Calculation of BytHrs again serves as a good example. We expect that for a checkpoint, the provider will directly measure the storage space actually occupied, whereas, for a given checkpoint time , the consumer will mimic the process by adding (for PUT) and subtracting (for DELETE) to calculate the space, and as we discussed with respect to Fig. 5, discrepancies are possible. Issues raised by clauses 1 and 2 can be directly addressed by the providers. A provider should evaluate their accounting models by performing consumer side accounting experiments to reveal any shortcommings. In particular, we recommend that for services that go through several state transitions (like EC2), providers should explicitly give FSM based descriptions, and ensure, as much as possible, that their models do not rely on unobservable (to consumer) events for billing charge calculations. On the whole, for IaaS services, consumer side accounting appears quite feasible. Whether this applies to PaaS and SaaS remains

72

A. Mihoob, C. Molina–Jimenez, and S. Shrivastava

to be seen. Any discrepancies that get introduced unintentionally (e.g., due to non identical checkpoint times) can be resolved by consumers by careful examination of corresponding resource usage data from providers. Those that cannot be resolved would indicate errors on the side of consumers and/or providers leading to disputes.

References 1. Amazon: Amazon simple storage service. developer guide, API version 2006–03–01 (2006), http://wwww.amazon.com 2. Amazon: Amazon ec2 faqs (2011), aws.amazon.com/ec2/faqs 3. Amazon: Amazon ec2 pricing (2011), aws.amazon.com/ec2/pricing 4. Amazon: Amazon elastic compute cloud (amazon ec2) (2011), aws.amazon.com/EC2/ 5. Amazon: Amazon elastic compute cloud user guide (api version 2011–02–28) (2011), docs.amazonwebservices.com/AWSEC2/latest/UserGuide/ 6. Bhushan, B., Tschichholz, M., Leray, E., Donnelly, W.: Federated accounting: Service charging and billing in a business-to-business environment. In: Proc. 2001 IEEE/IFIP Int’l Symposium on Integrated Network Management VII, pp. 107–121 (2001) 7. den Bossche, R.V., Vanmechelen, K., Broeckhove, J.: Cost-optimal scheduling in hybrid iaas clouds for deadline constrained workloads. In: Proc. IEEE 3rd Int’l Conf. on Cloud Computing (Cloud 2010), pp. 228–235 (2010) 8. Deelman, E., Singh, G., Livny, M., Berriman, B., Good, J.: The cost of doing science on the cloud: The montage example. In: Proc. Int’l Conf. on High Performance Computing, Networking, Storage and Analysis, SC 2008 (2008) 9. Elmroth, E., Marquez, F.G., Henriksson, D., Ferrera, D.P.: Accounting and billing for federated cloud infrastructures. In: Proc. 8th Int’l Conf. on Grid and Cooperative Computing, Lanzhou, Gansu, China, August 27-28, pp. 268–275 (2009) 10. de Leastar, E., McGibney, J.: Flexible multi-service telecommunications accounting system. In: Proc. Int’l Network Conf., INC 2000 (2000) 11. Mihoob, A., Molina-Jimenez, C., Shrivastava, S.: A case for consumer–centric resource accounting models. In: Proc. IEEE 3rd Int’l Conf. on Cloud Computing (Cloud 2010), pp. 506–512 (2010) 12. Molina-Jimenez, C., Cook, N., Shrivastava, S.: On the feasibility of bilaterally agreed accounting of resource consumption. In: 1st Int’l Workshop on Enabling Service Business Ecosystems (ESBE 2008), Sydney, Australia, pp. 170–283 (2008) 13. Murty, J.: Programming Amazon Web Services. O’Reilly (2008) 14. Wang, H., Jing, Q., Chen, R., He, B., Qian, Z., Zhou, L.: Distributed systems meet economics: Pricing in the cloud. In: Proc. 2nd USENIX Workshop on Hot Topics in Cloud Computing, HotCloud 2010 (2010)

Privacy Aware Obfuscation Middleware for Mobile Jukebox Recommender Services Ahmed M. Elmisery and Dmitri Botvich Waterford Institute of Technology – WIT, Telecommunications Software and Systems Group – TSSG, Co. Waterford, Ireland {ael-misery,dbotvich}@tssg.org

Abstract. Mobile Jukebox is a service offered by mobile operators to their clients, such that subscribers can buy or download anywhere, anytime fulllength music tracks over the 3G Mobile networks. Unlike some music download services, the subscribers can reuse the selected tracks on their music players or computers. As the amount of online music grows rapidly, Jukebox providers employ automatic recommender service as an important tool for music listeners to find music that they will appreciate. On one hand, Jukebox recommender service recommend music based on users’ musical tastes and listening habits which reduces the browsing time for searching new songs and album releases. On the other hand, users care about the privacy of their preferences and individuals’ behaviors regarding the usage of recommender service. This work presents our efforts to design an agent based middle-ware that enables the end-user to use Jukebox recommender services without revealing his sensitive profile information to that service or any third party involved in this process. Our solution relies on a distributed multi-agent architecture involving local agents running on the end-user mobile phone and two stage obfuscation process used to conceal the local profiles of end-users with similar preferences. The first stage is done locally at the end user side but the second stage is done at remote nodes that can be donated by multiple noncolluding end users that requested the recommendations or third parties mashup service. All the communications between participants are done through anonymised network to hide their network identity. In this paper, we also provide a mobile jukebox network scenario and experimentation results. Keywords: privacy, clustering, mobile jukebox, recommender service, multiagent.

1

Introduction

Being music a very important thing in people’s life, music applications are present in every computer and over many mobile devices. For such reason, mobile operators offer Mobile Jukebox as a moderate price service to their clients, such that subscribers can buy or download full-length music tracks over 3G Mobile networks. Unlike some music download services, the user can reuse the selected tracks on their music players or portable music devices. T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 73–86, 2011. © IFIP International Federation for Information Processing 2011

74

A.M. Elmisery and D. Botvich

According to [1] the more services appear in future, the more demand for personalization services will be to fight against information overload and find information relevant to each user. Recommender services can be seen as a suitable solution to these problems as they customize the offered services according to unique and individual needs of each user. The Jukebox providers employ recommender services to reduce browsing time for music listeners as the amount of online music grows rapidly. Jukebox recommender service becomes an increasingly important tool for music listeners to easily find new songs or playlists that they will appreciate. Examples of available Jukebox services are Apple iTunes® and Last.fm®. Apple iTunes automatically generates a playlist of songs from the user’s library which is similar to the selected songs, While Last.fm builds a detailed profile for each user’s musical taste by recording details of the songs the user listens to either from internet radio stations or user’s computer or portable music devices. This information is transferred to Last.fm’s database via music player and the user’s profile data can be displayed on his profile page. Jukebox recommenders commonly use collaborative filtering (CF) techniques to recommend music based on the listening behaviors of other music listeners. The Jukebox recommender harness the “wisdom of the crowds” to recommend music. Even though they generate good recommendations there are still some problems like the cold-start problem, as a recommender needs a significant amount of data before it can generate appropriate recommendations. However the acquisition, storage and application of sensitive personal information cause privacy concerns for users. There are many things having an impact on the perception of privacy for users like what kind of information is collected, how the information is used and the degree of accessibility of the information by others. The authors in [1] have done an empirical research concerning privacy preferences and individuals’ behaviors regarding personalization in music recommender systems. They found out that information about the purpose of the disclosure, recipients of the information, and the degree of the information involved and the benefits users expect to gain from disclosing personal information are the main factors influencing disclosure behavior. Based on their questionnaire in [2], participants were more willing to disclose music preferences than their personality. Participants considered information about personality traits more personal and more sensitive information than preferences for music genres. Participants expressed worries about not knowing how their information will be used in the system and who gets access to their personal information. The sensitivity of information affects on the disclosure decision. The questionnaire also shows that some participants even consider what benefits they will gain from disclosing the information. Participants can be divided into two groups based on their disclosure behavior, depending on whether they want to disclose anonymously or including identity information. One important factor have an impact on people’s disclosure behavior is the security and privacy standards taken by the Jukebox providers. In this work, we proceed with our approach presented in [3-7] to build AMPR (i.e. acronym for agent based middleware for private recommendations) that allows endusers to receive useful recommendations without disclosing their real preferences to the service. In the following section we will describe some properties for AMPR:

Privacy Aware Obfuscation Middleware for Mobile Jukebox Recommender Services

75

1. AMPR is running as a multi-agent based middleware to support different types of clients either thin or thick. Moreover, this architecture enables smooth integration with wide range of existing recommender services 2. AMPR preserves the aggregates in the obfuscated profiles to maximize their utility in order to attain acceptable recommendations accuracy, which facilitate AMPR to work with different state-of art filtering algorithms. Extra overhead in computation and communication to be added in the recommendation process due to the two stage obfuscation process 3. AMPR employs two stage obfuscation process to conceal the user’s preferences in his profile. The real user profile doesn’t leave his mobile phone until it is properly desensitized and it is maintained encrypted with private password that is known only to the user. If the user doesn’t accept to be tracked by the recommender service using his network identity, AMPR hides his identity by routing the submission of his locally obfuscated profile through relaying nodes in an anonymous communication network before sending it to the recommender service. In the rest of this paper we will generically refer to songs and playlists as Items. In section 2, we describe some related work. Section 3 introduces recommender system for mobile jukebox service scenario landing AMPR. Section 4 introduces our proposed solution. Section 5 describes the recommendation strategy used in our PCRS. Section 6 presents some experiments and results based on our proposed solution. Section 7 includes conclusions and future work.

2

Related Work

The majority of the literature addresses the problem of privacy for recommender services based on collaborative filtering technique, Due to it is a potential source of leakage of private information shared by the users as shown in [8]. In [9] it is proposed a theoretical framework to preserve the privacy of customers and the commercial interests of merchants. Their system is a hybrid recommender that uses secure two party protocols and public key infrastructure to achieve the desired goals. In [10, 11] it is proposed a privacy preserving approach based on peer to peer techniques using users’ communities, where the community will have a aggregate user profile representing the group as whole and not individual users. Personal information will be encrypted and the communication will be between individual users and not servers. Thus, the recommendations will be generated at client side. In [12, 13] it is suggest another method for privacy preserving on centralized recommender systems by adding uncertainty to the data by using a randomized perturbation technique while attempting to make sure that necessary statistical aggregates such as mean don’t get disturbed much. Hence, the server has no knowledge about true values of individual rating profiles for each user. They demonstrate that this method does not decrease essentially the obtained accuracy of the results. Recent research work [14, 15] pointed out that these techniques don’t provide levels of privacy as it was previously thought. In [15] it is pointed out that arbitrary randomization is not safe because it is easy to breach the privacy protection it offers. They proposed a random matrix based spectral filtering techniques to

76

A.M. Elmisery and D. Botvich

recover the original data from perturbed data. Their experiments revealed that in many cases random perturbation techniques preserve very little privacy. Similar limitations were detailed in [14].

3

Recommender System for Mobile Jukebox Service - Scenario

We consider the scenario where a private centralized recommender service (PCRS) is implemented on an external third party server and end-users give information about their preferences to that server in order to receive music recommendations. The user preferences stored in his profile in the form of ratings or votes for different item, such that items are rated explicitly or implicitly on a scale from 1 to 5. An item with rating of 1 indicates that the user dislikes it while a rating of 5 means that the user likes it. PCRS collects and stores different users’ preferences in order to generate useful recommendations. In this scenario there are two possible ways for user’s discloser: through his personal preferences included in his profile [16] or through the user’s network address (IP). AMPR employs two principles to eliminate these two disclosure channels, respectively. The obfuscation agents perturb user’s preferences for different items in his profile and the synchronize agent hides the user’s network identity by routing the

Fig. 1. Jukebox Service for Mobile Users with Third Party Private Recommender Service

Privacy Aware Obfuscation Middleware for Mobile Jukebox Recommender Services

77

communication with other participants through relaying nodes in Tor [17] anonymous network. The main challenge for synchronize agent is to tune up Tor and optimize its performance while maintaining the user anonymity. We don’t assume the server to be completely malicious. This is a realistic assumption because the service provider needs to accomplish some business goals and increase its revenues. In our framework, we will use the mobile phone storage to store the user profile. On the other hand, the Jukebox service maintains a centralized rating database that is used by the PCRS; Figure (1) shows the architecture of our approach. Additionally, we alleviate the user’s identity problems stated above by using anonymous pseudonyms identities for users.

4

Proposed Solution

In the next sub-sections, we will present our proposed middleware for protecting the privacy of users’ profiles.

Fig. 2. AMPR Components

Figure (2) demonstrates AMPR components that are running in the mobile phone at the user side. As shown, AMPR consists of different co-operative agents, A Learning agent captures user preferences about items explicitly or implicitly to build a rating table and meta-data table. The local obfuscation agent implements CTA obfuscation algorithm to achieve user privacy while sharing the data with other users or the system. The global perturbation agent is only invoked if the user is acting as a target user in recommendation process; it executes EVS-algorithm on the collected profiles finally. The synchronize agent is responsible for selecting the best suitable routing paths in the anonymised network to enhance its performance. The recommendation process based on the two stage obfuscation process in our framework can be summarized as following: 1. The learning agent collects user’s preferences about different items which represent a local profile. The local profile is stored in two databases, the first one is the rating database that contains (item_id, rating) and the other one is the metadata

78

2.

3.

4.

5.

A.M. Elmisery and D. Botvich

database that contains the feature vector for each item (item_id, feature1, feature2, feature3). The feature vector can include genre, author, album, decade, vocalness, singer, instruments, number of reproductions and so on. The target user broadcast a message to other users near him to request recommendations for specific genre or category of items. Individual users who decide to respond to that request use their local obfuscation agent to obfuscate a part of their local profiles that match query. The group members submit their locally obfuscated profiles to the requester using an anonymised network like TOR to hide their network identities. Enhancing the performance of communication through Tor is discussed in the next sub-section. If the size of group formation less than a specific value, the target user contacts the PCRS directly to gets recommendation from the centralized profiles stored in it. In order to hide items identifiers and meta-data from the requester and the PCRS. The manger agent at each participant side use locality-sensitive hashing (LSH) [18] to hash these values. One interesting property for LSH is that similar items will be hashed to the same value with high probability. PCRS is still be able to perform computations on the hashed items using appropriate distance metrics like hamming distance or dice coefficient. After the target user receives all the participants’ profiles (group profile), he/she incites his global perturbation agent to perturb the collected profiles. Then he can interact with PCRS by acting as an end-user has group profile as his own profile. The target user submits his/her group profile through an anonymised network to PCRS in order to attain recommendations. PCRS performs its filtering techniques on the group profile which in turn return a list of items that are correlated with that profile. This list is encrypted with a private key provided by target-user and it is sent back on the reverse path to the target user that in turn decrypts and publishes it anonymously to the other users that participated in the recommendation process.

4.1

Enhancing the Anonymized Network (Tor) Performance

Tor [17] is an anonymity network based on the original onion routing design but with several modifications in terms of security, efficiency, and deployability. The Tor network includes a small set of trusted authoritative directory servers responsible for aggregating and distributing signed information about known routers in the network. Tor clients periodically fetch the directory information from directory mirrors in order to learn information about other servers in the network. Tor network suffers from serious performance degradation because of its random path selection algorithms that use self-reported bandwidth values only, which might select with high probability a router with low bandwidth because it is sensitive to loads and changing network conditions. The synchronize agent seeks to enhance the performance by partitioning the Tor network into classes of high or low bandwidth Tor routers to better understand the relationships between different classes of routers /potential paths. Paths drawn from the class of high-bandwidth routers can provide better performance. Paths can be reserved for participating in specific recommendation

Privacy Aware Obfuscation Middleware for Mobile Jukebox Recommender Services

79

requests based on a user’s priorities or preferences. Therefore, inspired from the work in [19] we have implemented a simple parameterized path selection algorithm (PPS) that allow the synchronize agent to enhance the path selection in the Tor network with two priorities and it can be easily extended to support priorities larger than two. The synchronize agent can create circuits in advance to reduce the waiting time then measure the path throughput PT before using each path. PPS consists of the following steps: 1. The user input minimum path throughput PT , and circuit throughput CT to the synchronize agent. 2. Based on Tor authoritative directory servers, the algorithm start partitioning the class of high-bandwidth routers into a set of overlapping clusters (based on geographical location, platform, bandwidth, uptime, last update, number of connections and self-reported bandwidth estimate) using the algorithm on [20]. 3. It builds a pool of Tor nodes whose bandwidth from each cluster. In order to decrease the delay in circuit creation, the synchronize agent can select overlapping routers between clusters. 4. Then, it randomly builds circuits passing through these clusters. Then measure each circuit throughput, and select the first circuit that achieve bandwidth CT . 5. The synchronize agent then negotiates session keys with each router in the circuit. The exit router is responsible for establishing the connection from the Tor network to the client’s intended destination. 6. The synchronize agent records the previously used Tor nodes and exclude them from future circuit building clusters. 4.2

Proposed Obfuscation Algorithms

In the next subsections, we present two different algorithms used by the obfuscation agents in AMPR to obfuscate the user profile in a way that secure user’s preferences in PCRS with minimum loss of accuracy. Local Obfuscation using Clustering Transformation Algorithm (CTA). We proposed a novel algorithm for obfuscating the user profile before sharing it with other users. CTA designed especially for the sparse data problem we have here. CTA partitions the user profile into smaller clusters and then pre-process each cluster such that the distances inside the same cluster will maintained in its obfuscated version. We use local learning analysis (LLA) clustering method proposed in [21] to partition the dataset. After complete the partitioning, we embed each cluster into a random dimension space so the sensitive ratings will be protected. Then the resulting cluster will be rotated randomly. In such a way, CTA obfuscates the data inside user profile while preserving the distances between the data points to provide high accurate results when performing recommendations. The algorithm consists of the following steps: 1. The user ratings is stored in his mobile phone as dataset where each row is a sequence of attributes where

consists of rows, ……… .

80

A.M. Elmisery and D. Botvich

2. The dataset is portioned vertically into …… subsets of length , if / is not perfectly divisible then CTA randomly selects attributes already assigned to any subset and joins them to the attributes of the incomplete subsets. 3. Cluster each subset Using LLA algorithm, that result in clusters , , … . for each subset. So every point in the original dataset D falls exactly in one cluster. The aim of this step is to increase the privacy level of the transformation process and make reconstruction attacks difficult. 4. CTA generates two sets for each cluster in the subset these are and .Where is the set of points with highest values for field function and is the rest of construct a weighted graph Γ that contains its points in .For each point k-nearest neighbours in , each edge e Γ has a weight equals to the influence function of that point . 5. Estimate the geodesic distances by Computing the shortest distance between each two points in graph Γ using Dijkstra or Floyd algorithm and then build a distance . matrix 6. Based on ,we find a d-dim embedding space using classical MDS [22] as follows • Calculate the matrix of squared distances S D and the centering matrix H 1 1⁄N eeT • The characteristic vectors are chosen to minimize E τ D τ D L , where τ D is the distance matrix for the d-dim embedding space, and converts distances to inner products τ HSH/2. C , CTA randomly select two attributes x and x to 7. For each cluster perform rotation perturbation on selected attributes R x , x using transformation matrix M setup by the user for each cluster using range of angles defined in advance by the user. 8. Repeating steps 4-7 for all clusters in D to get the obfuscated portion D . Finally, the obfuscated dataset is obtained by D D. Global Perturbation using Enhanced Value-Substitution (EVS) Algorithm. After executing the local obfuscation process the global perturbation phase starts. The key idea for EVS is based on the work in [23] that uses Hilbert curve to maintain the association between different dimensions. In this subsection, we extend this idea as following, we also use Hilbert curve to map m-dimensional profile to 1-dimensional profile then EVS discovers the distribution of that1-dimensional profile. Finally, we perform perturbation based on that distribution in such a way to preserve the profile range. The steps for EVS algorithm consists of the following steps: 1. We denote the collected m-dimensional user profiles as dataset D of c rows, where each row is a sequence of m dimensions A A , A , A , A … … , A . 2. EVS divides the m-dimensional profile into grids of order k (where k is user defined value) as shown in [23, 24]. For order k, the range for each dimension divided into 2 intervals.

Privacy Aware Obfuscation Middleware for Mobile Jukebox Recommender Services

81

3. For each dimension of the collected profile : • Compute the k-order Hilbert value for each data point a . This value represents the index of the corresponding interval where it falls in. • EVS sort the Hilbert values from smallest to biggest, then use the step length (a user defined parameter) to measure whether any two values are near from each other or not. If these values are near, they are placed in the . same partition These two steps iterates for all m-dimensions. The final result from these steps is . partitions for each dimension denoted as 4. EVS constructs a N shared nearest neighbour sets S where r 1 … . N as in [25] from different partitions with a new modified similarity function as following, two partitions in different dimensions , form a shared nearest neighbour set S if they share k-number of common elements such that = . 5. For each newly created set S , EVS calculates the interquartile range. Then, for each point a S generate a uniform distributed random point in that range that can substitutes a . 6. Finally, the new set S is sent to PCRS.

5

Recommendation Strategy

PCRS employ online mode filtering algorithms to make predictions on the ratings of a particular user by collecting preference information from other users. The collected profiles (group profile) represented as user item matrix which contains a collection of numerical obfuscated ratings of M users on N items. After that, the neighbourhood formation at PCRS is done by calculating the similarity between users in the user-item matrix. Users similar to the target user using some proximity metric will form a proximity based neighbourhood with him [12]. This neighbourhood will utilize later for predication step. The prediction on rating of user i for item K is given by a weighted average [26] of users whose ratings are similar to the target user. P

∑

v

U

s u ,u

∑

v

v

s u ,u

U

Where U i U|v is the set of users who have rated the k th item. v is the mean of all ratings made by user i. The weights of average s u , u are the similarity between user u and u such as the Pearson correlation coefficient or Euclidean distance. We represent the user as a vector consists of n features slots, one for each item. These slots contain user’s ratings for different items or . The similarity between users’ vectors is calculated as the cosine of the angle formed between them as following: ∑

s u ,u v

v v v

v

v

82

A.M. Elmisery and D. Botvich

Finally, the recommendation process is to produce a predicted rating based on the neighbourhood for a list of items that have not been rated by the user, these items have a high potential of interest (predicated with high positive rating) to the user. The resulting items list can be further constrained based on marketing or Qos rules.

6

Experiments

The proposed algorithms are implemented in C++. We used message passing interface (MPI) for a distributed memory implementation of EVS algorithm to mimic a distributed reliable network of peers. We evaluated the proposed algorithms from two different aspects: privacy achieved and accuracy of results. The experiments presented here were conducted using the Movielens dataset provided by Grouplens [27]. The dataset contains users’ ratings on movies using discrete value between 1 and 5. The data in our experiments consists of 100.000 ratings for 1.682 items by 943 users. The experiments involve dividing the data set into a training set and testing set. The training set is obfuscated then used as a database for the PCRS. Each rating record in the testing set is divided into rated items and unrated items. The rated items are presented to the PCRS for making predication for the unrated items. To evaluate the accuracy of generated predications, we used the mean absolute error (MAE) metric proposed in [28]. MAE measures the predication verity between the predicated ratings and the real ratings, so smaller MAE means better recommendations provided by PCRS. To measure the privacy or distortion level achieved using our algorithms, we use variation of information metric VI [29] to estimate data error. Where, the higher VI means the larger distortion between the obfuscated and original dataset, which means higher privacy level. To evaluate the accuracy of CTA algorithm with respect to different number of dimensions in user profile, we control d-dim parameters of CTA to vary number of dimensions during the evaluation. Figure (3) shows the performance of recommendations of locally obfuscated data, as shown the accuracy of recommendations based on obfuscated data is little bit low when d-dim is low. But at a certain number of dimensions (500), the accuracy of recommendations of obfuscated data is nearly equal to the accuracy obtained using original data.

Fig. 3. Accuracy of recommendations for obfuscated dataset using CTA

Privacy Aware Obfuscation Middleware for Mobile Jukebox Recommender Services

83

In the second experiment performed on CTA algorithm, we examine the effect of ddim on VI values. As shown in Figure (4), VI values decrease with respect to the increase in d-dim values in user profile. d-dim is the key element for privacy level where smaller d-dim value, the higher VI values (privacy level) of CTA. However, clearly the highest privacy is at d-dim=100. There is a noticeable drop of VI values when we change d-dim from 300 to 600.d-dim value 400 is considered as a critical point for the privacy.Note that rotation transformation adds extra privacy layer to the data and in the same time maintains the distance between data points to enable PCRS to build accurate recommendation models. In the first experiment performed on EVS algorithm, we measured the relation between different Hilbert curve parameters (order and step length) on the accuracy and privacy levels attained. We map the locally obfuscated dataset to Hilbert values using order 3, 6 and 9. We gradually increased the step length from 10 to 80. Figure (5) shows the accuracy of recommendations based on different step length and curve order. We can see that as the order increases, the obfuscated data can offer better predictions for the ratings. This is because as the order has higher value, the granularity of the Hilbert curve becomes finer. So, the mapped values can preserve the data distribution of the original dataset. On the other hand, selecting larger step length increases MAE values as large partitions are formed with higher range to generate random values from it, such that these random values substitute real values in the dataset.

Fig. 4. Privacy levels for the obfucated dataset using CTA

Fig. 5. Accuracy level for different step length and orders for EVS

As shown in Figure (6), when the order increases a smaller range is calculated within each partition which introduces less substituted values compared with lower orders that attain higher VI values. The reason for this is larger order divides the mdimensional profile into more grids, which makes Hilbert curve better reflects the data distribution. Also, we can see that for the same Hilbert curve order the VI values are generally the same for different step length except for order 3, in which VI values has a sharp increase when step length grows from 50 to 60.The effect of increasing step length on VI values is more sensible in lower curve orders as fewer girds are formed and the increase of step length covers more portions of them, which will introduce a

84

A.M. Elmisery and D. D Botvich

higher range to generate raandom values from it. So the target user should select E EVS parameters in such a way to o achieve a trade off between privacy and accuracy. Finally, we measured th he overall performance of PPS algorithm in terms of the enhancement achieved in uploading time for the collected profiles. Figure (7) illustrates the cumulative distribution function (CDF) of time uploading the colleccted profiles of 331,25 bytes fro om the target user under the proposed PPS algorithm. T The term PPS = CT 34KB/s refers r to executing PPS with a circuit throughput equaal to 34KB/s. Table (1) gives th he mean, median for execution of PPS with different circcuit throughput values. Our anaalysis of Figure (7) and Table (1) lead us to the follow wing observations; the perform mance of Tor’s default path selection algorithm is unacceptable for responsivee recommender services. The largest uploading time for the profiles is 182.61s; also our o PPS algorithm significantly improves path selecttion performance. Table 1. 1 Uploading Time for Different CT values default

PPS= ۱‫ ܂‬൑10KB/s

PPS= ۱‫ ܂‬൑30KB/s

PPS= ۱‫ ܂‬൑34KB/s

Mean

30.38

25.34

19.64

12.54

8.41

Median

33.56

17.35

15.18

10.73

9.59

Fig. 6. Privacy level for diffeerent step length and orders for EVS

7

PPS= ۱‫ ܂‬൑20KB/s

Fig. 7. Uploading time using PPS Algorithhm

Conclusions and d Future Work

In this paper, we presented our ongoing work on building an agent based middlew ware for private recommendation services. We gave a brief overview of the recommendations process with w application to Jukebox music recommendations. A Also we presented the novel algo orithms that provide the users with complete control off the privacy of their profiles using two stage obfuscation process. We tested T The performance of these prop posed algorithms on real dataset. The experiential results show that preserving userss’ privacy for Jukebox recommender service is possibble.

Privacy Aware Obfuscation Middleware for Mobile Jukebox Recommender Services

85

In particular mean average error can be reduced with proper tuning of the algorithms’ parameters for large number of users. We realized that there are many challenges in building an agent based middleware scenario. This allow us to move forward in building an integrated system while studying issues such as a dynamic data release at a later stage and deferring certain issues such as virtualized schema and auditing to future research agenda. We need to perform extensive experiments in other real data set from the UCI repository and compare the performance with other techniques. Also we need to consider different data partitioning techniques as well as identify potential threats and add some protocols to ensure the privacy of the data against those threats. Acknowledgments. This work has received support from the Higher Education Authority in Ireland under the PRTLI Cycle 4 programme, in the FutureComm project (Serving Society: Management of Future Communications Networks and Services).

References 1. Perik, E., de Ruyter, B., Markopoulos, P., Eggen, B.: The Sensitivities of User Profile Information in Music Recommender Systems. In: Proceedings of Private, Security, Trust (2004) 2. Perik, E., de Ruyter, B., Markopoulos, P.: Privacy & Personalization: Preliminary Results of an Empirical Study of Disclosure Behavior. In: Proceedings of PEP, Edinburgh, UK (2005) 3. Elmisery, A.M., Botvich, D.: An Agent Based Middleware for Privacy Aware Recommender Systems in IPTV Networks. In: Watada, J., Phillips-Wren, G., Jain, L.C., Howlett, R.J. (eds.) Intelligent Decision Technologies. SIST, vol. 10, pp. 821–832. Springer, Heidelberg (2011) 4. Elmisery, A., Botvich, D.: Agent Based Middleware for Private Data Mashup in IPTV Recommender Services. In: 16th IEEE International Workshop on Computer Aided Modeling, Analysis and Design of Communication Links and Networks. IEEE, Kyoto (2011) 5. Elmisery, A., Botvich, D.: Agent Based Middleware for Maintaining User Privacy in IPTV Recommender Services. In: 3rd International ICST Conference on Security and Privacy in Mobile Information and Communication Systems. ICST, Aalborg (2011) 6. Elmisery, A., Botvich, D.: Privacy Aware Recommender Service for IPTV Networks. In: 5th FTRA/IEEE International Conference on Multimedia and Ubiquitous Engineering. IEEE, Crete (2011) 7. Elmisery, A., Botvich, D.: Private Recommendation Service For IPTV System. In: 12th IFIP/IEEE International Symposium on Integrated Network Management. IEEE, Dublin (2011) 8. McSherry, F., Mironov, I.: Differentially private recommender systems: building privacy into the net. In: Proceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 627–636. ACM, Paris (2009) 9. Esma, A.: Experimental Demonstration of a Hybrid Privacy-Preserving Recommender System. In: Gilles, B., Jose, M.F., Flavien Serge Mani, O., Zbigniew, R. (eds.), pp. 161–170 (2008) 10. Canny, J.: Collaborative filtering with privacy via factor analysis. In: Proceedings of the 25th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pp. 238–245. ACM, Tampere (2002)

86

A.M. Elmisery and D. Botvich

11. Canny, J.: Collaborative Filtering with Privacy. In: Proceedings of the 2002 IEEE Symposium on Security and Privacy, p. 45. IEEE Computer Society (2002) 12. Polat, H., Du, W.: Privacy-Preserving Collaborative Filtering Using Randomized Perturbation Techniques. In: Proceedings of the Third IEEE International Conference on Data Mining, p. 625. IEEE Computer Society (2003) 13. Polat, H., Du, W.: SVD-based collaborative filtering with privacy. In: Proceedings of the 2005 ACM Symposium on Applied Computing, pp. 791–795. ACM, Santa Fe (2005) 14. Huang, Z., Du, W., Chen, B.: Deriving private information from randomized data. In: Proceedings of the 2005 ACM SIGMOD International Conference on Management of Data, pp. 37–48. ACM, Baltimore (2005) 15. Kargupta, H., Datta, S., Wang, Q., Sivakumar, K.: On the Privacy Preserving Properties of Random Data Perturbation Techniques. In: Proceedings of the Third IEEE International Conference on Data Mining, p. 99. IEEE Computer Society (2003) 16. Parameswaran, R., Blough, D.M.: Privacy preserving data obfuscation for inherently clustered data. Int. J. Inf. Comput. Secur. 2, 4–26 (2008) 17. Dingledine, R., Mathewson, N., Syverson, P.: Tor: the second-generation onion router. In: Proceedings of the 13th Conference on USENIX Security Symposium, vol. 13, pp. 21–21. USENIX Association, San Diego (2004) 18. Indyk, P., Motwani, R.: Approximate nearest neighbors: towards removing the curse of dimensionality. In: Proceedings of the Thirtieth Annual ACM Symposium on Theory of Computing, pp. 604–613. ACM, Dallas (1998) 19. Pingley, A., Yu, W., Zhang, N., Fu, X., Zhao, W.: CAP: A Context-Aware Privacy Protection System for Location-Based Services. In: Proceedings of the 2009 29th IEEE International Conference on Distributed Computing Systems, pp. 49–57. IEEE Computer Society (2009) 20. Fellows, M.R., Guo, J., Komusiewicz, C., Niedermeier, R., Uhlmann, J.: Graph-Based Data Clustering with Overlaps. In: Ngo, H.Q. (ed.) COCOON 2009. LNCS, vol. 5609, pp. 516–526. Springer, Heidelberg (2009) 21. Elmisery, A., Huaiguo, F.: Privacy Preserving Distributed Learning Clustering Of HealthCare Data Using Cryptography Protocols. In: 34th IEEE Annual International Computer Software and Applications Workshops, Seoul, South Korea (2010) 22. Borg, I., Groenen, P.J.F.: Modern Multidimensional Scaling: Theory and Applications. Springer Series in Statistics. Springer, Heidelberg (2005) 23. Ghinita, G., Kalnis, P., Skiadopoulos, S.: PRIVE: anonymous location-based queries in distributed mobile systems. In: Proceedings of the 16th International Conference on World Wide Web, pp. 371–380. ACM, Banff (2007) 24. Reaz, A., Raouf, B.: A Scalable Peer-to-peer Protocol Enabling Efficient and Flexible Search (2010) 25. Jarvis, R.A., Patrick, E.A.: Clustering Using a Similarity Measure Based on Shared Near Neighbors. IEEE Trans. Comput. 22, 1025–1034 (1973) 26. Konstan, J., Miller, B., Maltz, D., Herlocker, J., Gordon, L., Riedl, J.: GroupLens: Applying Collaborative Filtering to {Usenet} News. Communications of the ACM 40, 77–87 (1997) 27. Lam, S., Herlocker, J.: MovieLens Data Sets. Department of Computer Science and Engineering at the University of Minnesota (2006) 28. Herlocker, J.L., Konstan, J.A., Terveen, L.G., Riedl, J.T.: Evaluating collaborative filtering recommender systems. ACM Trans. Inf. Syst. 22, 5–53 (2004) 29. Kingsford, C.: Information Theory Notes (2009)

Self-service Systems Performance Evaluation and Improvement Model Darius Dilijonas and Virgilijus Sakalauskas Department of Informatics, Vilnius University, Muitines 8, 44280 Kaunas, Lithuania {virgilijus.sakalauskas,darius.dilijonas}@khf.vu.lt

Abstract. The paper analyses the topic of service system productivity and profitability. Main focus of the research is self-service area, namely, the increase of ATM network productivity. Paper presents performance evaluation of self-service systems and improvement model for its increasing profitability. This model combines internal and external quality criteria and provides detailed understanding of the main components of productivity evaluation and methods. Using the model it is possible to create evaluation and improvement tools for increasing productivity of self-service systems. Experimental result shows that using the developed productivity model, ANN method and optimization procedure, productivity of ATM cash management could be increased by approximately 33 percent. Keywords: Self-service, ATM, productivity, profitability, ANN.

1

Introduction

Globalization and market liberalization imply changes in basic economic understanding. Organizations begin to realize that the essential issue is not just profit maximization but profit maximization by increasing consumers’ satisfaction with the provided services. At present, not just the delivery of services is important; more attention is given to the way of providing the services. Development of information technologies significantly impacts the render of services and the entire sector of services. Consumers now have more information and better understanding of any service, therefore, the competition between service providers is determined by quality, and quality determines profitability [1]. Gartner [2] notices, that there have been major changes in the service sector. Traders are beginning to realize that technologies may have critical importance for the increase of profit and operational efficiency. The paper analyses the topic of service system productivity and profitability. The work focuses on self-service area, namely, the increase of ATM network productivity. The profitability of electronic service systems depends on their productivity and quality of the services provided. It is possible to ensure the productivity of service systems by efficient planning and by improving delivery process. In order to avoid affecting delivery process in service systems, it is very important to make timely decisions. It is impossible to store services like goods; therefore, inadequate decisions are related to the direct loss of income and quality decrease.

T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 87–98, 2011. © IFIP International Federation for Information Processing 2011

88

D. Dilijonas and V. Sakalauskas

The productivity of electronic services may be ensured by implementing advanced real-time decision management systems and by changing the realization of service system quality from traditional to predictive. The quality of such systems is ensured by predicting service system behavior. Multi-agent technologies are used for data collection from distributed service points [3]. Intelligent agents evaluate the demand of service system resources according to the collected data, and plan the work to have sufficient resources for services and to maintain high quality and maximum system availability. Applying intelligent systems of resource planning and optimization, the work of service system must be managed in real time. Certainly, without service quality and productivity management models and systems, the mentioned measures are only the tools; therefore performance management models should be applied as well. The goal of the research is to perform evaluation of self-service systems and to develop improvement model for increasing its profitability. Service system profitability increases when improves the quality of provided services and decrease supply costs. Due to insufficient level of their computerization, the developed method may be hardly applied for rendering traditional services. Therefore, more restricted area of service systems is analyzed in the paper - self-service facility systems – ATM networks. The second section of the paper presents analysis of the literature concerning service systems productivity. The self-service systems performance evaluation model is described in the third section. The forecasting and optimization methods are described in the fourth section. According to presented evaluation model, fifth section provides detailed description of experimental results of ATM networks profitability simulation.

2

Service Systems Productivity

The concept of productivity in service area started to be analyzed only at the end of the XX century. Service sector is one of the fastest growing economy segments [4]; still only few empirical surveys have been conducted in productivity management area. Gronroosa [5], Chase and Haynes [6] structured the elements of service productivity theory. They analyzed the similarities and differences of production and service productivity, and formulated the main definitions of service productivity. Ojasalo [7] structured the model of service productivity. Sherman [8] defined service productivity components. Seth [1] carried out survey on quality management models, developed in the period of 1984-2003. It is obvious that service delivery process changes from traditional to the one reasoned by information technologies. Application of information technologies in service quality management is becoming highly important. Service sector productivity depends on technologies and process automation measures [9], at present this tendency is even more significant because the competition in service sector is increasing. Service delivery through self-service terminals or computer systems allows increasing the efficiency of service system and the quality of providing services. It is obvious that higher profitability is typical for efficiently operating systems [8]. Semeijn and others [10] showed that the work of servicing system is none the less important than the e-service system itself. Implementation of the best and most functional ATMs cannot ensure high level of self-service facilities without an appropriately functioning service system. Voss’s [11] survey shows that service quality expectations in retail banking only approximately fulfill the service received.

Self-service Systems Performance Evaluation and Improvement Model

89

Problems, which appear in service systems, may disturb delivery process [12]; in most cases unavailable services greatly reduce the loyalty of a service [13]. For this reason it is important to combine internal and external quality criteria. Only one or another group of criteria is usually analyzed in literature, and a common model of self-service network quality is not provided. Productivity evaluation system of e-services is discussed by Gronroos [5]; Ojasalo, [7]; Rutkauskas et al. [14]. Service productivity management is a very important topic, but very little empirical research has been carried out in this area [15]. Productivity consists of 4 components [8]: price, resource allocation, technological and scale efficiency. The main dilemma of productivity is balancing of revenue and costs [16]. Depending on demand, it is necessary to ensure efficient use of resources because services cannot be stored. Productivity is divided into internal and external. Internal productivity describes how efficiently resources are used. External productivity shows what service profitability is. Services productivity is the balance of external and internal efficiency [6]. Another important element of productivity is demand management or capacity efficiency, because supplier cannot store services [5]. Capacity efficiency shows how efficiently system internal resources satisfy the demand. The process of services is an open system therefore application of traditional efficiency evaluation systems may determine false results. Productivity may be evaluated totally or partially. According to Gronroos [5], the only theoretically and practically significant way to calculate services productivity is financial measures. Internal quality of a system is ensured by using advanced management tools, that allow optimizing operational costs and service delivery process, and by using internal quality management systems, which are based on sets of productivity criteria. Clearly detailed productivity criteria of ATM network internal quality management systems are analyzed in articles of Aldlaigan and Buttle [17], Bahia and Nantel [18], Jabnoun and Al-Tamimi [19], Joseph and Stone [20]. In order to ensure external quality, process development measures and external quality management systems are necessary. External quality management systems were analyzed by Lovelock [21], Johnston and Clark [22], Edvardsson and others [23], they settled value-based service quality management models, which analyze economic and social service delivery aspects.

3

Self-service Systems Performance Evaluation

The main quality evaluation criteria of self-service systems were analyzed by Aldlaigan and Buttle [17], Bahia and Nantel [18], Jabnoun and Al-Tamimi [19], Joseph and Stone [20]. Most authors distinguish 5 criteria, which make the basics of ATM service quality: sufficient size of ATM network, safe environment, userfriendly connection, convenient allocation, and ATM functionality. Evaluation of service delivery should be carried out according to the criteria of e-service benefit/value: environmental features, security, communication speed, reliability, customer support, responsiveness, information completeness, availability, delivery and personalization [24]. The figure 1 shows the model of self-service system performance criteria; the model elaborates productivity components and sets out methodologies, which may be applied to ensure internal, external and capacity efficiency. It was found that internal

90

D. Dilijonas and V. Sakalauskas

productivity of self-service terminals may be increased by using cheaper maintenance resources if it does not have negative impact on quality, also by implementing more advanced technological decisions, and by optimal distribution of resources. Service resources may be optimized the way that their amount in self-service infrastructure would always satisfy the demand, therefore it is necessary to know what the demand will be, and forecast tools used for its determination. Knowing the demand, it is possible to foresee necessary amount of resources, and this is especially important in such service systems, where resources are of short validity or are realizable and can be used to earn income by another way.

Fig. 1. Self-service system performance criteria model

Productivity is comprised of several components that independently influence the overall efficiency of the organization. The components are 1) price efficiency, 2) allocative efficiency, 3) technical efficiency, and 4) scale efficiency [8]. Price efficiency requires purchasing the inputs that meet the quality standard at the lowest price. Allocative efficiency is the use of the optimal mix of inputs to produce the products or services. Technical inefficiency exists when it is possible to produce more outputs with the inputs used or where it is possible to produce the actual outputs with fewer inputs. Scale efficiency is the component that addresses the optimal activity volume level, producing more or less goods or services than the optimal level results in added costs solely due to volume or size [8].

Self-service Systems Performance Evaluation and Improvement Model

91

Different management tools and methodologies can be used to evaluate and manage service productivity. In model, these tools are classified by three categories: K – Qualitative, S – Quantitative and without marking both K/S. The model line connecting the methodology and the performance component shows which technique is applicable to evaluate performance of specific component. The external productivity consists of profitability and quality evaluation criteria’s. Self-service Systems Performance Assessment Model Productivity may be evaluated by physical, financial and combined measures; they may totally or partially evaluate productivity [7]. All productivity evaluation criteria may be derived from them (Table 1), in accordance with the self-service systems performance criteria model. Table 1. Self-service systems productivity evaluation criteria’s Evaluation criteria Maximum Capacity = work time / average duration of one transaction System Load = Number of transactions performed / work time | maximum limitation under the load condition “Freeze” Resources = returned or lost resources / all resources | rating per terminal Availability = no working time / total hours worked | rating per terminal Reliability = failed transactions / total transactions | rating per terminal Total Maximum Load = (work time / average time of one transactions) * number of terminals | maximum limitation under the overall load condition; Total Load = number of transactions performed / total amount of resources; Total “Freeze” Resources = returned or lost resources / all resources | rating for all terminals Total Availability = no working time / total hours worked | rating for all terminals Total Reliability = failed transactions / total transactions | rating for all terminals

1

1

Features FIZ

2

Component MEF

3

Type DP

FIZ

MEF

DP

FIZ

MEF

DP

FIZ

TNAS

DP

FIZ

TNAS

DP

FIZ

MEF

PP

FIZ

MEF

PP

FIZ

MEF

PP

FIZ

TNAS

PP

FIZ

TNAS

PP

Features: FIZ - physical, FIN - financial, KOM - combined; Component: KNAS - Price efficiency; IPNAS - Allocative efficiency; TNAS - Technical efficiency ; MEF - Scale efficiency; PEL – Profitability; 3 Type: PP – total productivity; DP - partially productivity; 2

92

D. Dilijonas and V. Sakalauskas Table 1. (Continued)

Support Performance = [revenue of services = (transaction price + sold product)] / labor costs | rating for all terminals; Service delivery efficiency = [revenue of service = (transaction price + sold product)] / all costs | | rating for all terminals; Terminals Revenue Productivity = revenue of services = (transaction price + sold product) / number of terminals Terminal Cost Efficiency = number of transactions performed / resource costs Productivity = number of transactions or sold product units / (outsourcing, depreciation, maintenance, support or services delivery costs)

FIN

KNAS

DP

FIN

KNAS

PP

KOM

PEL

DP

KOM

IPNAS

PP

KOM

KNAS

DP

DEA analysis [8] is most suitable for evaluation of internal performance of selfservice systems. Self-service system productivity evaluation model (all criteria)

Define (D)

Selected Performance Model

Criteria 1 Criteria 2 ... Criteria n

Measure (M)

Criteria value

Analyze (A)

Causes

Implement (I)

Performance improvement goals

Control (C)

Comparison before and after changes

Data Envelopment Analysis (DEA) Indicator analysis process analysis, staff models. Other methods

Fig. 2. The model of self-service system performance assessment

Electronic services quality model is applied for the evaluation of external performance of self-service systems [25]. This model provides full set of criteria for the assessment of service content and functionality. IT-based model [26], which assesses IT importance for the delivery of services, is applied for evaluation of social aspects: age, mood for using e-services. The figure 2 presents self-service system performance model. Self-service system performance assessment model needs to be formulated in accordance with the criteria,

Self-service Systems Performance Evaluation and Improvement Model

93

formulated in the table 1. First the calculations are prepared and then the most appropriate evaluation criteria are selected. Following this, the productivity model is developed and simulations consisting real data are done. The obtained results are evaluated by comparing the results between ATM’s. After analysis of the results, the reasons for low productivity are identified. Eventually tools have to be prepared to eliminate the causes of low productivity. The goals are set for performance improvement. After the changes and improvement are made, the comparison has to be done, in order to evaluate whether the realized changes increased productivity.

4

Profitability Simulation

This section describes experimental results of ATM network profitability analysis, using proposed performance evaluation methods and models. 4.1

Artificial Neural Network Model of Demand Forecasting

A flexible neural network model has been created, which changes its structure depending on situation and is used to forecast self-service facility demand. Considering the forecast results, the optimization procedure foresees optimal resource loads of ATM network. In our past researches we have already presented experimental research of ANN model [27]. We came to the conclusion, that the best forecasting results are reached using direct spread multilayer neural network with one hidden layer, consisting of fifteen neurons in the hidden layer (transfer function – hyperbolic tangent) and one output (linear neuron), chosen for ATM cash prediction. Neural network has 6 inputs: average cash demand of the last 7 days, day of the week, month of the year, day of the month, days before holidays and time series record number. The output is ATM cash demand forecast for the following day or week if week discretion data is transferred. Flexibility of neural network is controlled by limiting the weight of neural networks. For this reason, additional member D is introduced into the neural network prediction error expression, which punishes the network for high weight values (i.e. too high flexibility). 4.2

ATM Network Optimization of Resource Delivery

ATM network optimization topics include three areas: selection of optimal ATM network size [29], cash demand specifics (size of cash on the market) [30-33], and forecast of cash flow demand in ATM networks [27,28]. Analysis of applied methods showed that most frequently analyzed economic models (dependencies) influence cash market structure by factors. Real decisions and tools, which could help to optimize ATM cash management, are not provided. For profitability modeling we are using ATM optimization function to model such amounts of cash load, which minimize ATM operation expenses. Optimization is performed in consideration of interest rate, upload costs, forecast recommendations

94

D. Dilijonas and V. Sakalauskas

and cash management limitations: cash balance, cash insurance expenses, maximum amount of cash, minimal amount of cash, possible ATM unavailability period. Output of optimization function is the date of cash upload into ATM and optimal cash amount, which is necessary to upload in an ATM in order to minimize ATM operation expenses. ATM cash management costs = Cash * Interest + Cash * Insurance + (RUC or EUC) + Penalty → minimize, where

(1)

Cash - ATM cash amount (limitations: possible minimum and maximum load); Interest - interest rate, %; Insurance - insurance rate, %; RUC - Regular upload costs; EUC - Extra upload costs; Penalty – penalty if ATM is not accessible more than 2% of all working time. 4.3

ATM Networks Profitability Simulation

Experimental research of 2 types has been carried out in the paper: low load and high load ATM network profitability simulation. High load ATM networks are replenished more than one time per day, via versa low load. For high load ATM network simulation we used India bank data of 5500 ATMs (duration up to 3 months). For low load ATM network simulation we used data of Lithuanian bank of 21 ATM (duration - 6 months). In both cases we evaluated optimal (when ANN method is used) and typical (real bank process) models. After modeling high load ATM network productivity, we found that average productivity of ANN model for a group of accurate ATMs is 17.73 % (assessing losses 11.16%). Optimal model, compared to a typical one is 18% more productive. This criterion characterizes capacity productivity component and shows how productively the existing resources may be used. It is attributable to financial productivity criteria, and only partially evaluates productivity. The amounts of cashback are decreased by 90%. This criterion shows the productivity of resource distribution. ATM model average productivity for a group of average ATMs is 26.27% (assessing losses 12.18%), amounts of cash-back are decreased by 32.2% (assessing loss 11.16%), and the amounts of cash-back are reduced by 653%. Benefit without loss is the highest in the group of inaccurate ATMs, its increase is determined by obvious facts, i.e., bigger amounts of cash are uploaded in hardly forecasted ATMs, because standard deviation is much higher than in the case of accurate and average group models. In regard to accurate ATM group, a network of 5000 ATMs can save 4250 Euros per day, and if the free cash is lent, 2660 Euros may be earned. Total saving and profit benefit would reach 6900 Euros per day, what would allow making 2.52 million Euro profit per year. Average profit of all model groups is 2.35 million Euros per year. Cash management costs are approximately reduced by 25%. After modeling low load ATM network productivity (detailed evaluation results are presented in Tables 2 and 3), we found that in optimal case average cash-back amounted 13.15%; in typical case – 37.23%.

Self-service Systems Performance Evaluation and Improvement Model

95

Table 2. ATM network management efficiency modeling results of the Lithuania Bank (Optimal Model)

Optimal model (using ANN forecasting and optimization procedures) ATM number ATM5011 ATM1004 ATM1045 ATM2001 From 2009-5 2009-2 2009-3 2009-5 Till 2010-2 2009-11 2010-2 2010-2 Average load 73928 58752 35732 36574 Average Cash back 4225 5462 8277 3687 Cash Back % 5.71% 9.30% 23.17% 10.08% Cash Load Quantity 50 46 36 35 Cash Price (interest from 244 400 723 281 cash back) Cash Load Price 1225 1125 875 850 Total Cash Price (CB) 1469 1525 1598 1131 Frozen Cash (interest) 1469 2188 1709 1457 Total Cash Price (UZ) 3409 3313 2584 2307 Efficiency (CB) 47.11% 59.77% -2.51% 68.88% Efficiency (UZ) 44.75% 28.20% 28.36% 45.90%

ATM7012 2009-5 2010-2 39695 6938 17.48% 31 460 750 1210 1264 2014 28.80% 15.18%

Table 3. ATM network management efficiency modeling results of the Lithuania Bank (Typical Model)

Typical model (typical bank scenario) ATM number ATM5011 ATM1004 ATM1045 From 2009-5 2009-2 2009-3 Till 2010-2 2009-11 2010-3 Average load 147616 88533 58685 Average Cash back 27505 49540 14313 Cash Back % 18.63% 55.96% 24.39% Cash Load Quantity 38 49 23 Cash Price (interest 1827 2566 984 from cash back) Cash Load Price 950 1225 575 Total Cash Price (CB) 2777 3791 1559 Frozen Cash (interest) 5220 3390 3031 Total Cash Price (UZ) 6170 4615 3606

ATM2001 ATM7012 2009-5 2009-5 2010-2 2010-2 46321 41401 26190 12686 56.54% 30.64% 59 34 2158 850 1475 3633 2789 4264

850 1700 1525 2375

Using the created ATM network cash management model, the amounts of cashback were decreased by 24.08%, compared to a real scenario. Model efficiency shows that using the created ANN method and optimization procedure, the cash supply productivity of certain ATMs may be increased by 44.75% (ATM5011), 45.90% (ATM2001).

96

D. Dilijonas and V. Sakalauskas

Figure 3 presents graphic generalization of modeling results of ATM No 5011; we notice that in case of optimal scenario, cash collecting is balanced in respect of cash demand. Using the created ANN method and optimization procedure, cash supply productivity of ATM5011 is increased by 44.75% (ATM5011), compared to the model, applied in a bank.

Fig. 3. Cash load model of ATM5011

Using the created ATM network cash management model for low load ATM network management, the cash-back amounts are reduced by approximately 24%, compared to a real scenario. Using the created ANN method and optimization procedure, ATM cash management performance may be approximately increased by 33 %.

5

Conclusions

The productivity of electronic services may be ensured by implementing advanced real-time decision management systems, i.e., changing the realization of service system quality from traditional to predictive. Application of information technologies in service quality management is becoming highly important. The work of service system must be managed in real time, applying intelligent systems of resource planning and optimization. Certainly, without service quality, productivity management models and systems, the advance software is only a tool; therefore performance management models should be applied as well.

Self-service Systems Performance Evaluation and Improvement Model

97

It was found that internal productivity of self-service terminals may be increased by using cheaper maintenance resources, by implementing more advanced technological decisions, and, by optimal distribution of resources. More productive distribution of resources may be performed using accurate service demand forecast measures. Analysis of high and low load ATM network profitability showed that proposed self-service systems performance evaluation and improvement methods allow decreasing the amounts of cash-back by 24% and increasing cash management productivity up to 33%.

References 1. Seth, N., Deshmukh, S.G., Vrat, P.: Service quality models: a review. International Journal of Quality & Reliability Management 22(9), 913–949 (2005) 2. Gartner: Emerging Technologies Will Help Drive Mainstream BI Adoption (2008), http://www.gartner.com/ 3. Dilijonas, D., Bastina, L.: Retail banking optimization system based on multi-agents technology. In: 16th WSEAS International Conference on Computational Intelligence, Man-Machine Systems and Cybernetics, December 14-16, pp. 204–209 (2007) 4. Sahay, B.S.: Multi-factor productivity measurement model for service organization. International Journal of Productivity and Performance Management 2(1), 7–22 (2005) 5. Gronroos, C., Ojasalob, K.: Service productivity Towards a conceptualization of the transformation of inputs into economic results in services. Journal of Business Research 57, 414–423 (2004) 6. Chase, R.B., Haynes, R.M.: Service operations management. A field guide. In: Swartz, T.A., Iacobucci, D. (eds.) Handbook of Services Marketing and Management, pp. 455–472. Sage Publications, Thousand Oaks (2000) 7. Ojasalo, K.: Conceptualizing productivity in services. Hanken Swedish School of Economics Finland/CERS Center for Relationship Marketing and Service Management. Helsinki/Helsingfors (1999) 8. Sherman, H.D., Zhu, J.: Service Productivity Management, Improving Service Performance using Data Envelopment Analysis (DEA), pp. 2–5. Springer, Heidelberg (2006) 9. Gummesson, E.: Productivity, quality and relationship marketing in service operations. International Journal of Contemporary Hospitality Management 10(1), 4–15 (1998) 10. Semeijn, J., Van Riel, A.C.R., Birgelen, M.J.H., Streukens, S.: E-services and offline fulfillment: how e-loyalty is created. Managing Service Quality 15(2), 182–194 (2005) 11. Voss, C.A.: Rethinking paradigms of service – service in a virtual environment. International Journal of Operations & Production Management 23(1), 88–105 (2003) 12. Zhang, X., Prybutok, V.R.: A consumer perspective of e-service quality. IEEE Transactions on Engineering Management 52(4), 461–477 (2005) 13. Watcher, K.: Longitudinal assessment of web retailers: issues form a consumer point of view. Journal of Fashion Marketing and Management 18(1), 53–69 (2002) 14. Rutkauskas, J., Paulaviciene, E.: Concept of Productivity in Service Sector. Engineering Economics 3(43), 29–34 (2005); Influence of quality management of the country’s economy, ISSN 1392-2785

98

D. Dilijonas and V. Sakalauskas

15. Johnston, R.: Service productivity: Towards understanding the relationship between operational and customer productivity / R. Johnston, P. Jones. International Journal of Productivity and Performance Management 53(3), 201–213 (2004) 16. Gronroos: Service management and marketing. A customer relationship management approach. Wiley, Chichester (2000) 17. Aldlaigan, A., Buttle, A.: SYSTRA-SQ: a new measure of banks service quality. International Journal of Service Industry Management 13(4), 362–381 (2002) 18. Bahia, K., Nantel, J.: A reliable and valid measurement scale for the perceived service quality of banks. The International Journal of Bank Marketing 18(2), 84–91 (2000) 19. Jabnoun, N., Al-Tamimi, H.A.: Measuring perceived service quality at UAE commercial banks. International Journal of Quality & Reliability Management 20(4), 458–472 (2003) 20. Joseph, M., Stone, G.: An empirical evaluation of US bank customer perceptions of the impact of technology in service delivery in the banking sector. International Journal of Retail & Distribution Management 31(4), 190–202 (2003) 21. Lovelock, C.H.: Functional integration in service: understanding the links between marketing, operations, and human resources. In: Swartz, T.A., Iacobucci, D. (eds.) Handbook of Marketing and Management. Sage, Thousand Oaks (2000) 22. Johnston, R., Clark, G.: Service Operations Management. Prentice Hall, London (2001) 23. Edvardsson, B., Enquist, B., Hay, M.: Values based service brands: narratives from IKEA. Managing Service Quality 16(3), 230–246 (2006) 24. Rowley, J.: An analysis of the e-service literature: towards a research agenda. Internet Research 16(3), 339–359 (2006) 25. Santos, J.: E-service quality: a model of virtual service quality dimensions. Managing Service Quality 13(3), 233–246 (2003) 26. Zhu, F.X., Wymer, W.J., Chen, I.: IT-based services and service quality in consumer banking. International Journal of Service Industry Management 13(1), 69–90 (2002) 27. Simutis, R., Dilijonas, D., Bastina, L.: Cash demand forecasting for ATM using neural networks and support vector regression algorithms. In: Continuous Optimization and Knowledge-Based Technologies, 20th EURO Mini Conference, May 20-23, pp. 416–421 (2008) 28. Sakalauskas, V., Dilijonas, D., Kriksciuniene, D., Simutis, R.: Intelligent systems for retail banking optimization-Optimization and Management of ATM Network System. In: ICEIS 2009 - 11th International Conference on Enterprise Information Systems, Milan, vol. AIDSS, pp. 321–325 (2009) 29. McAndrews, J., Rob, R.: Shared ownership and pricing in a network switch. International Journal of Industrial Organization 14, 727–745 (1996) 30. Boeschoten, W.: Cash management, payment patterns and the demand for money. De Economist 146(1), 117–142 (1998) 31. Viren, M.: Financial innovations and currency demand, some new evidence. Empirical Economics, 451–461 (1992) 32. Snellman, H., Viren, M.: ATM Networks and Cash Usage. Fore presentation at the FMG/Cass Business School Workshop on Financial Regulation and Payment Systems at the Cass Business School on September 28 (2006) 33. Stix, H.: How do debit cards affect cash demand? Working Paper 82, Oesterreichische National bank (2003)

Service Provisioning Framework for Digital Smart Home Services Egidijus Kazanavicius1, Darius Imbrasas2, and Mantas Razukas2 1

Real Time Computing Systems Centre, Kaunas University of Technology, Studentu St. 50-213, Kaunas, Lithuania [email protected] 2 Private Joint Stock Company "Elsis TS", Uosio St. 10, Kaunas, Lithuania {darius.imbrasas,mantas.razukas}@elsis.lt

Abstract. “Development and Management Framework of Smart Home Services (SNAPAS)” is an innovative system, which seeks to fill the market niche of Smart Home services and fully complies with the priorities of scientific development approved by the European Parliament and Council. Framework is characterized by an open component-based pluggable architecture, which provides new forms of interactive services in home environments, including the third-party service providers in the business model. SNAPAS is intended to develop, provide and control the services of Smart Home. Statistical research shows that the most important groups of Smart Home services are comfort, energy efficiency, and security. The most important are the energy efficiency services, because 76.3 % of all respondents would like to improve their heating system. Keywords: smart home, service provisioning, distributed systems.

1

Introduction

At the present time the systems and technologies of Smart Home are rapidly growing in popularity around the world. These systems include the automation of various housing areas such as accounting and control of consumed energy and resources, control of lighting and climate, security control, digital multimedia home systems, automated supervision of home environment and many others. The increasing demand for such systems is in the countries of both North and South America, in Asia, Australia, Europe and Russia. Development programs of Smart Home are even included in the priority fields of research in European Union countries. According to the 7th Framework Program, approved by European Parliament and by the European Council, research is divided into six specific programs, including “Collaboration” which includes the trend of information and communication technologies. The strategic goal of this trend is “To increase competitiveness of European industry and to provide the opportunity for Europe to manage and shape the future development of information and communication technologies (ICT) in such a way that it would satisfy the needs of its T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 99–109, 2011. © IFIP International Federation for Information Processing 2011

100

E. Kazanavicius, D. Imbrasas, and M. Razukas

society and economy. Activity will consolidate the base of European science and technology, will ensure its global leadership in ICT, will help to manage and stimulate innovations through ICT and will ensure that ICT progress was rapidly transformed into benefits for European citizens, business and industrial companies and government authorities”. In this context “Development and Management Framework of Smart Home Services (SNAPAS)” is an innovative system, which seeks to fill the market niche of Smart Home services and fully complies with the priorities of scientific development approved by the European Parliament and Council, seeking to ensure competitiveness of Lithuania and Europe, satisfying the needs of its society and economy. “Development and Management Framework of Smart Home Services” is characterized by an open component-based pluggable architecture, which provides new forms of interactive services in home environments, including the third-party service providers in the business model (which is not yet in action worldwide).

2

Methodology for Smart Home Services Provisioning

“Development and Management Framework of Smart Home Services” is intended to develop, provide and control the services of Smart Home. JSC “Elsis TS” is in the center of this system chain of service development and provision, i.e. it is the developer, provider and manager of Smart Home services, whereas end-users of Smart Home services are Smart Home residents. The provider of Smart Home services will allow its customers to select different home services and will effectively manage the relationship between the customer-user and the providers of third party services. SNAPAS is intended for several user groups. System users are grouped into four groups: end-users of Smart Home services; developers of Smart Home services; providers of Smart Home services; managers of Smart Home services. Developers of Smart Home services are various organizations that are interested in providing their services to residents, using “Development and Management Framework of Smart Home Services”. Any company can be the developer of Smart Home services, provided that a tripartite agreement between the company, the service provider and the service manager is signed. Service developers can potentially be the organizations that already provide a range of services to residents by other means (not digital): companies providing energy resources, banks, operators of telecommunications, enterprises for distribution of information, television and radio operators, medical institutions, etc. The system allows service developers to develop, deploy and test their services remotely, via “on-line” mode and in real time, using special user friendly interfaces. The providers of Smart Home services will ensure system performance for the provision of services, by providing hosting. The group of service providers may consist of both Internet service and resource providers. These users will be able to monitor and manage the quality of services in real time. The managers of Smart Home services will provide supervision, installation, activation and deactivation of provided services. They will be able to perform all control operations remotely from the service center, which also receives

Service Provisioning Framework for Digital Smart Home Services

101

information about system state in its separate parts in real time, also from homes of Smart Home’s services’ users. The system will automatically notify about various emergency situations, their cause and location. Thus the system SNAPAS is designed to provide the end-user with innovative services from anywhere and at any time, by fully automating the process which is intended to develop and provision services of home automation, ensuring integration of these services with already existing systems and services. The system will ensure control and monitoring of services in real time and via “on-line” mode from anywhere and at any time, and also monitoring and control of service quality, and detection of emergency situations. 2.1

Scenario for Smart Home Services Provisioning

Fig. 1 shows the scenario of development, control and provisioning of Smart Home services: 1. Service developers (which can be third party service developers) develop the services of Smart Home using “Development and Management Framework of Smart Home Services” by local or remote means. When developing a new service the developer has the possibility to test that service in the test environment of the SNAPAS system. 2. After the new service is created, service managers validate this service i.e. they check if this newly developed service meets all requirements of quality, security, content and others. Having ensured that the service meets these requirements, the service manager will make the final service configuration, setting the model of service provisioning (the payment model, the distribution model, etc.). 3. The newly developed, validated and configured service is automatically installed into service centers of residential homes or directly into residential homes remotely (via Internet or other) in accordance with the distribution model. In the case when residential houses belong to a block of residential houses with a neighborhood service center, services are automatically installed through that center. Otherwise, if there is no service center, the service is installed directly on the server at the end-user’s home. 4. The new service is automatically integrated into the package of pre-existing services. The end-user can use this service free of charge for a period of time and if this service meets the customer’s needs then it can be subscribed to interactively, using any terminal equipment (TV, PC, mobile devices, etc.) with a graphical interface. “Development and Management Framework of Smart Home Services” develops and integrates different types of services, such as: 1.

Comfort related services: lighting control, ventilation control, heating control, control of shutters, control of security systems, etc. 2. Entertainment services: media centre, digital TV, video rental (VoD), games, etc.

102

E. Kazanavicius, D. Imbrasas, and M. Razukas

3. Residential environment services: home residents’ health care services, health care services for elderly and disabled residents, services for environmental monitoring and determination of risk factors. 4. Services of supervision centers: tracking and control services of house engineering equipment, security services, Internet security services, heating control services (of the residential block), etc. 5. Other services whose need may arise in the future or services developed by third parties.

Fig. 1. Scenario for development, control and provisioning of Smart Home services by JSC “Elsis TS”

SNAPAS features open architecture, i.e. it allows you to add new Smart Home services in real time, without changing the existing system architecture and structure. Smart Home end-users can try out, select and control the desired services as well as unsubscribe at any time interactively in real time. Such a solution is innovative in the world context, because until now global markets offer inflexible specialized solutions of certain companies, thus forcing the users to select standard services in advance without being able to easily abandon them later or subscribe to new services without changing the existing Smart Home system, which is costly and sometimes impossible. Currently offered commercial Smart Home solutions are usually closed and do not allow third-parties to provide services, however a new product is being planned on the

Service Provisioning Framework for Digital Smart Home Services

103

basis of which new business models will form, in which all interested third-parties will be allowed to provide Smart Home services on the basis of a pre-existing Smart Home infrastructure. A versatile service stack and structure is defined in a way that allows immediate integration of new services without changing the existing system, whose engineering services will include control and monitoring of engineering house systems, the provision of digital interactive content and control of residential environment.

3

Statistical Market Research

In order to assess the need and appeal to consumers of the system being developed, the evaluation of the Smart Home services system’s conceptual solution was performed by taking part in the exhibition Namu pasaulis 2010 (Home World 2010). Evaluation of the conceptual solution by taking part in the exhibition Namu pasaulis 2010 (Home World 2010) was chosen due to the fact that many people who are interested in housing issues (construction, renovation, other solutions and innovations intended for modern homes) come to that exhibition. In order to properly evaluate the conceptual solution of Smart Home services, analysis of the Smart Home services system’s concept was commissioned for the exhibition, which helped to discover the opinion of exhibition visitors about Smart Home services. In addition, the concept of the Smart Home service system was introduced to visitors and participants. The questionnaire of 16 questions was approved, drawn in accordance with rules of “ICC/ESOMAR International Code of Marketing and Social Research Practice” and in accordance with all quality standards of social studies. It helped to understand and clarify the views and needs of potential Smart Home consumers, which made it easier to select specific applied research activities. 430 people participated in the survey. Thus the distribution error of overall responses with probability of 95 percent does not exceed 4.69 % (based on sample calculation formulas of T.Yamane and V.A. Jadov). The survey was carried out in accordance with quality standards of the World Association for Public Opinion Research (WAPOR) (WAPOR Code of Professional Ethics and Practices). At the beginning of the survey subjects were asked whether they have heard of systems such as “smart home“, “future home”, “home automation”, ”intellectual home environment”, etc. 70 % of respondents indicated that they heard of these systems, but could not specify how they function. 18.8 % claimed that they had not heard about such systems and did not know what it is and 11.2 % indicated that they were aware of these systems and had a lot of information about it. Participants were asked to indicate the most important issue in terms of their housing. They were asked to lay out answers in order of priority. Slightly more than half of survey participants (50.9 %) indicated that convenience and comfort is an essential feature in terms of their housing. 25.8 % claimed that their priority is energy saving. 19.5 % claimed that the most important feature is security, and 3.7 % said introduction/possession of new, modern technologies in their homes.

104

E. Kazanavicius, D. Imbrasas, and M. Razukas

Fig. 2. Distribution of responses to the question “What is the most important to you in terms of your home“, in percentage, N=430

Subjects were also asked to indicate the most important home security service. Almost half of respondents (47.7 %) claimed that the most important is security against break-ins, and 38.8 % of respondents said fire safety. 7.7 % of respondents claimed that the key factor for them is protection against gas leakage, and 6 % said protection against water leakage. 86 % of survey participants indicated that heating control is/would be important for them. Control of lighting systems and other electrical appliances would be important for 63.3 % of respondents, control of ventilation/air conditioning is important for 50.5 % of respondents, and data acquisition from accounting devices on water, gas and electricity consumption as well as data submission is important for 39.3 %. The study participants were also asked to give their opinion on how much energy they believe they can save using an energy regulation system. About half of survey participants claimed that it is possible to save up to 30 % of energy resources if an energy regulation system is used. Almost one third of respondents (30.9 %) believe that an energy regulation system can save up to 10 % of energy resources and 10.9 % of respondents think that they can save up to 50 % of energy resources. Some respondents (7.9 %) believe that an energy regulation system can save up to 5 % of energy resources. According to the survey, 76.3 % of respondents claim that they would like to improve the heating system most of all in order to reduce the cost of consumed energy. 10.7 % of respondents would like to improve the lighting system most of all, 5.1 % of respondents claimed they would like to improve the ventilation system and the same percentage of respondents (5.1 %) would like to improve the air conditioning system in order to reduce costs. 2.8 % of respondents said that they would not like to improve any of the above systems in their homes. The study shows that more than one third of all respondents (35.3 percent) would like to remotely manage the security of their homes, i.e. they would like to receive notifications of burglary, fire, water, gas leaks, etc. 28.6 % of respondents said that they would like to control their home micro-climate (heating, ventilation/air

Service Provisioning Framework for Digital Smart Home Services

105

Fig. 3. Distribution of responses to the question “Which system would you like to improve in your home in order to reduce the cost of energy used“, in percentage, N=430

conditioning) remotely in their homes, 21.9 % said that most of all they would like to control lighting and other electrical devices remotely in their homes. 10.9 % would like to control the home environment (yard, garage doors, lawn watering) remotely and 3.3 % of respondents said that they would not like to control any of these functions remotely in their homes. Respondents were also asked to indicate the most convenient way to control the above mentioned services. Those who said that they would not like to control any of these functions remotely were asked to imagine the most convenient way to remotely control any of these services. 42.8 % of respondents said that the most convenient way to control these services would be by telephone. 40.9 % of respondents indicated by computer as one of the most convenient way to control services remotely.

Fig. 4. Distribution of responses to the question “Which functions would you like to control remotely in your home“, in percentage, N=430

106

E. Kazanavicius, D. Imbrasas, and M. Razukas

38.6 % of respondents indicated the remote control as one of the most convenient tools. According to the survey the least popular tool for remote service control is a menu on the TV screen.

Fig. 5. Distribution of responses to the question “Which is the most convenient way for you to control the above mentioned services“, in percentage, N=430

For 70.5 % of respondents, modern technologies in housing are mostly associated with automated control of housing and its devices. For 15.8 % of respondents modern technologies in housing are mostly associated with robots, performing a variety of household chores. For 10.2 % of respondents modern technologies in housing are mostly associated with the latest audio/video systems, and for 3.5 % of respondents with house/apartment decoration according to the modernist style (beginning of twentieth century- middle of twentieth century ). 46.7 % of respondents indicated that the most relevant service for them is automated control of lighting when being away from home (the simulation of someone being at home). 42.6 % stated that they would like to have a modern door, allowing people to enter the house without a key, displaying the people who are waiting at the door and allowing entry of guests while being anywhere in the house/flat. 40.2 % of respondents would like to have voice control of electrical appliances. According to the survey the least relevant services for the respondents would be e-health services (interactive upload of health data, e.g. blood pressure, cardiogram, ultrasound, from your house to your doctor before visiting him). Only 15.3 % of respondents are interested in such services. 11.9 % of respondents claimed that they would not want to have any service mentioned in the questionnaire in their homes. 72.8 % of survey participants indicated that they would like to use Smart Home services in their homes. 20.7 % of respondents said that they still do not know whether they would like to use Smart Home services in their homes and 6.5 % claimed that they would not like to use such services in their homes.

Service Provisioning Framework for Digital Smart Home Services

107

Fig. 6. Distribution of responses to the question “Do you wish to use Smart Home services in your home? “, in percentage, N=430

3.1

Market Research Summary

Even 81.2 % of respondents have heard of schemes such as “smart home“, “future home”, “home automation”, ”intellectual home environment”, etc. and 11.2 % of them have a lot of information about it. The essential feature in terms of housing for slightly more than half of respondents is comfort, and for a quarter of the respondents is energy saving. Monitoring their own security burglar alarm system is/would be the most important for nearly half of respondents, and for 38.8 % it is fire safety. Heating control services is/would be important for 86 % of respondents. Control of lighting systems and other electrical appliances would be important for 63.3 % of respondents. Meteorological data (meteorological station) is the least relevant for all respondents (only 7.0 %). About half of survey participants believe that it is possible to save up to 30 % of energy resources if an energy regulation system is used and almost one third of respondents believe that an energy regulation system could save up to 10 % of energy resources. 76.3 % of respondents would like to improve the heating system most of all in order to reduce the cost of consumed energy. More than one third of all respondents (35.3 %) would like to remotely manage the security of their homes, i.e. they would like to receive notifications of burglary, fire, water, gas leaks, etc. Telephone, computer and remote control were named as the most popular measures to control services remotely. The least popular measure is by a menu on the TV screen. For 70.5 % of respondents modern technologies in housing is mostly associated with automated control of housing and its devices. From all systems that are not widely known in Lithuania but are already in demand abroad, the greatest interest would be in simulation of being at home, a modern door which allows people to enter the house without any key, showing the people waiting at the door and allowing guests to enter while being at any spot of the house/flat as well as voice control of electrical appliances. 11.9 % of respondents claimed that they would not want to have any service mentioned in the questionnaire. 72.8 % of survey participants indicated that they would like to use Smart Home services in their homes, 6.5 % would not like to use them and 20.7 % still do not know.

108

4

E. Kazanavicius, D. Imbrasas, and M. Razukas

Conclusions

The above described results showed that the greatest interest in Smart Home services is among people 18-44 years old, applied research must be carried out while developing system interfaces oriented to the largest potential consumer group. The most important groups of Smart Home services are as was previously thought but with different priorities: comfort, energy efficiency, security. This confirms our understanding as well as the trend of applied research activity. The sequence of the main comfort services shows that when developing the SNAPAS concept the major focus is on remote data reading from accounting devices – TMT must be developed in a direction allowing the Smart Home system to effectively integrate this type of service. Another issue that emphasizes the importance of applied research of energy efficiency services is the fact that 76.3 % of all respondents would like to improve their heating system. Thus this area appears to be a big focus. Applied research must be carried out while developing system interfaces oriented to the most important tools that control Smart Home services. These tools are a telephone, a computer and remote control. Due to the fact that almost 73 % of survey participants wish to use Smart Home services in their homes, we can confidently claim that we have chosen the correct avenue of applied research and our work is meaningful.

References 1. Zhenxing, W., Zhongyuan, L., Linxiang, S.: The Smart Home Controller Based on Zigbee. In: 2nd International Conference on Mechanical and Electronics Engineering (ICMEE), vol. 2, pp. V2-300–V2-302 (2010) 2. Xuemei, L., Gang, X., Li, L.: RFID Based Smart Home Architecture for improving lives. In: 2nd International Conference on Anti-counterfeiting, Security and Identification, ASID 2008, pp. 440–443 (2008) 3. Byoung-Kug, K., Sung-Kwa, H., Young-Sik, J., Doo-Seop, E.: The Study of Applying Sensor Networks to a Smart Home. In: Fourth International Conference on Networked Computing and Advanced Information Management, NCM 2008, vol. 1, pp. 676–681 (2008) 4. Holzner, C., Guger, C., Edlinger, G., Gronegress, C., Slater, M.: Virtual Smart Home Controlled By Thoughts. In: 18th IEEE International Workshops on Enabling Technologies: Infrastructures for Collaborative Enterprises, WETICE 2009, pp. 236–239 (2009) 5. Jahn, M., Jentsch, M., Prause, C.R., Pramudianto, F., Al-Akkad, A., Reiners, R.: The Energy Aware Smart Home. In: 5th International Conference on Future Information Technology (FutureTech), pp. 1–8 (2010) 6. Mineno, H., Kato, Y., Obata, K., Kuriyama, H., Abe, K., Ishikawa, N., Mizuno, T.: Adaptive Home/Building Energy Management System Using Heterogeneous Sensor/Actuator Networks. In: 7th IEEE Consumer Communications and Networking Conference (CCNC), pp. 1–5 (2010) 7. Al-Qutayri, M., Barada, H., Al-Mehairi, S., Nuaimi, J.: A Framework for an End- to-End Secure Wireless Smart Home System. In: 2nd Annual IEEE Systems Conference, pp. 1–7 (2008)

Service Provisioning Framework for Digital Smart Home Services

109

8. Pensas, H., Vanhala, J.: WSN Middleware for Existing Smart Homes. In: Fourth International Conference on Sensor Technologies and Applications (SENSORCOMM), pp. 74–79 (2010) 9. Papadopoulos, N., Meliones, A., Economou, D., Karras, I., Liverezas, I.: A Connected Home Platform and Development Framework for Smart Home Control Applications. In: 7th IEEE International Conference on Industrial Informatics, INDIN 2009, pp. 402–409 (2009)

Open-Source Software Implications in the Competitive Mobile Platforms Market Salman Qayyum Mian1, Jose Teixeira2, and Eija Koskivaara3 1

Nokia Siemens Networks (NSN), Linnoitustie 6, 02600 Espoo, Finland [email protected] 2 Turku Center for Computer Science (TUCS), Joukahaisenkatu 3-5 B, 20520 Turku, Finland [email protected] 3 Turku School of Economics (TSE), Rehtorinpellonkatu 3, 20500 Turku, Finland [email protected]

Abstract. The era of the PC platform left a legacy of competitive strategies for the future technologies to follow. However, this notion became more complicated, once the future grew out to be a present with huge bundle of innovative technologies, Internet capabilities, communication possibilities, and ease in life. A major step of moving from a product phone to a smart phone, eventually to a mobile device has created a new industry with humongous potential for further developments. The current mobile platform market is witnessing a platforms-war with big players such as Apple, Google, Nokia and Microsoft in a major role. An important aspect of today's mobile platform market is the contributions made through open source initiatives which promote innovation. This paper gives an insight into the open-source software strategies of the leading players and its implications on the market. It first gives a precise overview of the past leading to the current mobile platform market share state. Then it briefs about the open-source software components used and released by Apple, Google and Nokia platforms, leading to their mobile platform strategies with regard to open source. Finally, the paper assesses the situation from the point of view of communities of software developers complementing each platform. The authors identified relevant implications of the open-source phenomenon in the mobile-industry. Keywords: open-source, platform strategies, mobile industry, mobile platforms, iOS, Android, Symbian, Maemo.

1

Introduction

The open-source software phenomenon continues, persistently capturing the attention of both scholars and practitioners. It started in 1985, when Richard Stallman founded the Free Software Foundation promoting the idea of freedom in software. The Foundation, still very active today, promotes that software could run freely and that T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 110–128, 2011. © IFIP International Federation for Information Processing 2011

Open-Source Software Implications in the Competitive Mobile Platforms Market

111

correspondent software source code could be studied, changed, copied, published and also distributed freely. Raymond (2001) popularized the phenomenon by studying the development of the first free operating system known as GNU/Linux. It was claimed that Linus Torvalds steered a totally new way of developing software by making use of thousands of volunteer developers collaborating over Internet in a distributed “organization” towards a common goal. Representing a more efficient way than the traditional hierarchical and controlled way used by corporate software houses. The echo of open-source software development attracted an interest particularly from the economic scholars. Lerner and Tirole (2001) made a preliminary exploration of the economics of open-source by assessing the extent to which economics literature on “labor economics” and “industrial organization theory” could explain the opensource phenomenon. The mentioned research brought some answers on what motivates open-source developers, compared the different programming incentives between open-source and proprietary settings, and highlighted the favorable organizational and governance characteristics for open-source production. Many other relevant contributions followed Tirole’s research agenda: such as Paajanen (2007), that developed a multiple case study on the licensing for opensource software; and Bonaccorsi and Rossi (2003) that discussed the coexistence of open-source and proprietary software in organizations. Other studies give evidence to a large scale adoption of open-source software by organizations; perhaps one of the most clear evidences illustrating the growing economic impact of open-source comes from Finland. According to Helander et al. (2008), 75% of the studied Finnish firms were using open-source software, a enormous increase since only approximately 13% of the same firms were using open-source software according to an analogous survey conducted in 2000. The term platform is conceptually abstract and widely used across many fields. Within this research, the term platform maps the concept of computer-based platform as in Morris and Ferguson (1999), Bresnahan and Greenstein (1999) and West (2003). As mentioned by West (2003), a platform consists of an architecture of related standards, controlled by one or more sponsoring firms. The architectural standards typically encompass a processor, operating system (OS), associated peripherals, middle-ware, applications, etc. Platforms can be seen as systems of technologies that combine core components with complementary products and services habitually made by a variety of firms (complementers). The platform leader and its complementers jointly form an “ecosystem” for innovation, that increases platform's value and consequently user’s adoption (Gawer and Cusumano 2008). For instance, the current leaders of the video games industry, operate by developing the hardware consoles and its peripherals while providing a programmable software platform that allows complementers to develop games on top of their systems. Attracting more game developers to the platform means more and much better games, an increase in value for the end users (video game players).

112

S.Q. Mian, J. Teixeira, and E. Koskivaara

It is empirically observable that the current mobile industry is shifting from a products-war paradigm to a distinct platforms-war situation. Rather than vendors competing on a basis of its perceived product features, we observe a complex network of vendors under the umbrella of a platform aimed in becoming the “defacto” standard under a market subject to the forces of network externalities described by Shapiro and Varian (1999). This empowers the importance of platform complementers such as telecommunication operators, semi-conductors components makers, producers of hardware accessories, 3rd party software application developers, etc. As mentioned by West (2003), device makers differentiation might not be driven by higher architectural layers but by efforts in the systems integration and design. The future of the mobile platform market is as debatable as global warming issues, though the earlier inevitably presents a promising and better future. Just a look at the past ten years give an impression of a primitive era, which has grown out to be a huge bundle of innovative technologies, Internet capabilities, communication possibilities, and ease in life. It all started from a simple product phone, then to a smart phone but the technological developments forced the use of a more general term, such as mobile device. In the current mobile devices market, there are still doubts over what defines a successful mobile device, is it the hardware, or the operating system, or the ease of use. The success of a mobile device has become a combination of all these things including the brand name, open source code, variety of applications, and many more aspects. According to Eric Schmidt, Google’s executive chairman, brand is just a little part, the high-tech war between companies is about coming up with the ideas, revolution in applications, and the chance to lead in the future mobile devices industry (Zakaria, 2011). From the consumer’s point of view, the mobile devices market is moving very fast and is volatile. In a more general perspective, it has become a race to be the best operating system for the Internet enabled phones (Ocock, 2010). The current mobile device platforms market is mainly dominated by Google, Apple, and the one with its own legacy, Nokia. When it comes to platform, each of them promotes their own, Android from Google, iOS from Apple and the historical Symbian from Nokia. While Apple’s iOS is restricted to their own device iPhone, Symbian is mostly to Nokia, Motorola and Samsung phones especially the current version Symbian^3. Android from Google is the most versatile in terms of adoption, being used in mobile devices from other companies such as HTC, Samsung, and LG, to name a few. This research mainly focuses on Android, Symbian, Maemo and iOS platforms leaving the others like RIM, Windows mobile platform etc. This is mainly because other platforms either do not have a significant market share or they clearly state that their interests do not match with that of the open source community. Symbian retains the credit of being the oldest smart-phone platform in use and it corresponds to them being the biggest operating system by market share at the moment (Ocock, 2010). All of these platforms have market share varying according to geographical locations in the world. But in general others have captured the market share mostly from Symbian in the last few years. The Fig. 1 below shows the worldwide smart-phone market shares percentages by operating systems in the last two years, according to Gartner, 2011.

Open-Source Software Implications in the Competitive Mobile Platforms Market

113

Fig. 1. Worldwide smart-phone Market shares (%) by platform in 2009/2010 (Gartner, 2011)

The technological developments in various mobile device platforms has eventually introduced tough competition, with eventually consumer winning in the end. One of the adoptions on its way is Microsoft’s Windows phone 7 OS taken by Nokia, which is a strategic step taken by the company assessing the current market (Nokia press release, 2011). However, with increasing competition, the mobile devices industry has also been marred with lawsuits. In the recent years, the above mentioned supreme leaders have now and then been involved in various patents and copyright cases against each other. Another aspect of the current platforms market is the code being open source (meaning available to everybody), with the perception of achieving innovation and creativity by getting all the developers involved. However, among the above companies this positive initiative varies on different grounds. On the other hand, an Open Handset Alliance led by Google was founded in Nov, 2007 with the purpose of accelerating innovation in mobile and to richly improve the consumer experience. The alliance is a group of 84 technology and mobile companies which together released the Android with the aim of deploying handsets and services using the Android platform. The alliance is committed to great openness for the development of the Android platform through open software and applications (Open Handset Alliance, 2011). Considerable research was established on technological platform strategies, being briefly identified here: Anchordoguy (1989) exploited the rich competition between computer platforms in Japan while the western world was being monopolized by IBM. Bresnahan and Greenstein (1999) examined thirty years of the computer industry from a pure economical perspective. West (2003) investigated in detail, the hybrid strategies from PC vendors that attempted to combine the advantages of opensource software while keeping tight control and differentiation in its platforms.

114

S.Q. Mian, J. Teixeira, and E. Koskivaara

The Japanese PC and Gaming industry carefully reviewed by Hagiu (2004) introduced the concept of multi-sided platforms and surveying prevalent business models adopted by dominant platform makers. Gawer and Cusumano (2008) made key contributions by introducing the coring and tipping two strategic abstractions. However, with a stronger empirical relevance of open-source factors and with a completely new mobile industry in context, current established research needs to be further developed.

2

Methodology

As the title of this paper suggests, the research question of this study is “What are the open-source software implications in the competitive mobile platforms market?”. Forced by the magnitude of the research question, the authors fractionated the research problem with the following three research questions: First, “What are the OSS components integrated by Apple, Google and Nokia in their mobile platforms?”; Second, “What are the open-source platform-based strategies employed by Apple, Google and Nokia?”; And finally, “How are the 3rd party developers coping with the announced strategies”? The authors addressed the research questions with Yin (1991) case study research methodology. The research authors had tiny or no control over networked behavioral events within the complex market of mobile device platforms being studied. Moreover, this research focuses on contemporary phenomenon in a real-life context where the boundaries between phenomenon being studied and its context are not obvious. According Flyvbjerg (2006), an in-depth case study research approach provides a systematic way of looking at events, collecting data, analyzing information, and reporting valuable results as knowledge. Following Yin (1991) case study methodology our case is “implications of the open-source software in the mobile industry” and our units of analysis are the four mobile platforms developed by the Apple, Google and Nokia, giants of the telecommunication industry: “iOS”, “Android”, “Symbian” and “Maemo”. According to the same author’s taxonomy, our research is a multiple descriptive case study. It follows an embedded design with multiple units of analysis where the authors are looking for consistent patterns of evidence across the four units within the same phenomenon being studied. The mobile devices industry is recent, and the emergence of mobile devices platforms is ever newer. Perhaps the EPOC system that surged in mid 1996 empowering Psion devices is the first mobile device platform. Later known as Symbian, it provided differentiated structures for handset device makers and application developers, instantiating the concept of multi-sides platforms described by Hagiu (2005). The novelty of the phenomenon being studied constrained the work for the authors in finding early theoretical knowledge addressing the research problem. The researchers strongly believe that it is dangerous to generalize research from the previous two decades on the computer-based platforms, to mobile platforms: the market players are different, the technology and geographies too. With the lack of early knowledge the authors were forced to seek an exploratory case-study over

Open-Source Software Implications in the Competitive Mobile Platforms Market

115

descriptive or explanatory approaches. The authors share the same view as Flyvbjerg (2006), crediting that concrete and context-dependent knowledge is often more valuable that vain search for predictive theories and universals. The data was systematically collected between the 20th of May and 8th of April 2011, from a set of Internet sites which are mostly publicly available. One of the research authors subscribed the software development programs from Apple and Nokia to get access to information targeting each platform’s third party software developers. The following Table 1 presents the different websites from where the case study data was collected. The authors explored a key strength of the case study method by making use of multiple sources and techniques (Yin 2002). The systematically analyzed Internet sites greatly differ in vendor control, editorial constraints, pluralism and interactivity. The data sources can be categorized in four group-types with regard to developers’ goals and policies: First, press releases provided by the platform vendors were studied; followed by software development portals and discussion forums covering each of the platform, then generalist business, economical and technological press; and finally, websites with a very strong focus on reviewing the personal electronics industry. Table 1. Internet captured research data sources description Site

Description

Vendor controlled? http://www.apple.com/pr/ Apple press release Yes http://www.google.com/press/ Google press release Yes http://press.nokia.com/ Nokia press release Yes Yes. Partially http://developer.apple.com/devce Apple developers portal nter/ios

http://developer.android.com/res ources/community-groups.html http://www.forum.nokia.com/

Android developers Yes. Partially community groups Nokia interface with Yes. Partially its developers

http://maemo.org/

No Nokia Maemo platform community portoal Nokia Symbian Yes. platform portal

http://symbian.nokia.com/

http://www.businessweek.com

A weekly business magazine. With strong USA focus.

No

Confidentiality Public Public Public Public or constrained to 3rd party developers Public Public or constrained to 3rd party developers Public

Public or constrained to 3rd party developers Public

116

S.Q. Mian, J. Teixeira, and E. Koskivaara Table 1. (Continued)

http://www.wired.com/

http://news.cnet.com/

http://thsnews.com

http://ostatic.com/

http://globalpublicsquare.blogs.c nn.com/

http://www.zdnet.com/ http://slashdot.org/

http://www.engadget.com

http://eu.techcrunch.com/

Magazine covering how technology affects culture, the economy, and politics. Strong USA focus CNET provides reviews of both consumer electronics and software. Independent generalist news provider. Portal reviewing open-source software and services. Generalist blog feed by TIME and CNN journalists among other contributors. Technology news and product reviews Technology news forum hosting many discussion on opensource topics A web magazine with focus on consumer electronics. With strong USA focus. A blog based edition covering Web 2.0 and Mobile startups. EU focus.

No

Public

No

Public

No

Public

No

Public

No

Public

No

Public

No

Public

No

Public

No

Public

The data was collected and classified by taking into account the Romano et al. (2003) methodology for analyzing web-based qualitative data encompassing the three, elicitation, reduction and visualization processes. Data elicitation and reduction was performed over the Internet by collaborative manners, using the popular googledocs web-based software. Reduction included the grouping of data into different codes derived from previous theory and observed text data, such coding took into account some of the qualitative research principles stipulated by Seaman (1999). Both the

Open-Source Software Implications in the Competitive Mobile Platforms Market

117

popular office-suite package and the freeMind software application were intensively used during the visualization phases, allowing the authors to identify patterns and structures which permitted drawing of conclusions.

3

Findings and Research Agenda

Targeting the first research question, assessing what open-source software components are integrated by Apple, Google and Nokia in their mobile platforms, the researchers identified most of them from the official platform websites. These are enumerated in the following Table 2, in terms of the software packages resulting from the vendor integration of technological assets provided by the open-source community. Every package found was carefully investigated by the authors and rich data was collected for post-analysis. Collected evidence included, a brief description of each package, the correspondent open-source project website, the open-source license in which the software is distributed and, wherever possible, the organization behind the identified software project. Table 2. Platforms architecture reviewed data sources Platform iOS 4.3.3 iOS 4.3.3 Android 2.3 Maemo 4.1 Maemo 4.1

Website http://www.opensource.apple.com/release/ios-433 http://www.opensource.apple.com/release/developer-tools-40 http://android.git.kernel.org/platform/external/ https://garage.maemo.org/docman/view.php/106/354/maemopackages20080725.ods http://maemo.org/maemo_release_documentation/maemo4.1.x/

The software packages obtained from the list of websites in Table 2 were verified to present real world devices. By this, the researchers ensured that each package found was present in a physical device empowered by the studied platforms. However, due to technical and legal constraints, it was only possible to do such verification on the Maemo and Android platforms, using a Samsung Galaxy S mobile phone and a Nokia N810 internet tablet. All packages found were present in the real world physical devices. A considerable amount of open-source software packages were found, 28 packages within the iOS 4.3.3 platform, 108 packages within the Android 2.3 platform and 151 packages in the Maemo platform. During the data collection period, on 6th of April, Nokia announced a radical change on the organization strategy abandoning the Symbian platform’s open-source strategy and the corresponding Symbian foundation. The source-code is no longer available, and the official Nokia platform website at http://symbian.nokia.com read “Not open-source, just open for business”. And the platform website is not even available from non-official websites like http://sourceforge.net/projects/symbiandump. Since the Symbian operating systems did not had strong connections with the open-source community from the start, the

118

S.Q. Mian, J. Teixeira, and E. Koskivaara

authors decided to drop any investigation regarding what Symbian platform packages were derived from the open-source community. However it is important to note especially when addressing the second research question, that the last version of the Symbian platform included the open-source Qt technology. Described at http://qt.nokia.com, it allowed third party application developers to target different Nokia platforms, using a single software development kit. Even if Maemo version 4.1 was able to run applications developed in Qt, official support of Qt was only announced for the Maemo version 5. Nokia purchased Trolltech, the Norwegian company behind Qt, in January 2008 for $153 million and immediately started promoting a three platforms strategy, which included Symbian S40, Symbian S60 and Maemo, targeting different ranges of devices. But a unified software development kit based on Qt allowed developers to quickly port their applications for different Nokia platforms and devices as well. Much before the full implementation of the above mentioned three platforms strategy, the appointment of Stephen Elop as CEO on 21 September 2010 brought radical changes in Nokia strategy. The new CEO appointed the Windows Mobile platform as the new primary platform for Nokia, turning Microsoft from rival competitor to a strategic partner. On a curious note, Stephen Elop had moved to Nokia from a lead executive position at Microsoft. On February 2011, Nokia announced that the new Nokia Windows mobile platform will not support the Qt technology. This move caused strong dissatisfaction among Nokia third party software developers that they would need to completely re-develop their applications to a new platform with unknown capabilities. The open-source community behind Qt, with strong contributions from Bogdan Vatra quickly announced the port of the Qt core technology to the rival platform Android on March 2011. The link of Nokia with the Qt technologies ended in the same month with the partial sale of Nokia’s Qt business operations to Digia, a Finnish IT services provider. From the analysis of software packages resulting from the vendor integration of technological assets provided by the open-source community, we can conclude that Apple uses open-source software at low extend. iOS 4.3.3 makes extensive use of open-source components, but only in its operating system core, a layer completely hidden from the iPhone device users. The only exception is the web browsing technology which is based on the webKit open-source project available at http://webkit.org/. Apple is the main contributor to the webKit open-source project initiated as a fork of the KHTML open-source project from the KDE open-source community. It was noticed that Apple integrates older versions of open-source projects releases, perhaps seeking architectural simplicity and stability over the integration of the last project features, more prone to bugs. Regarding open-source components integrated by the Android 2.3 platform, it is visible that Google uses open-source components at large. We can induce that the Android platform uses many recent versions of open-source project releases and that many of the used open-source components are heavily modified to facilitate adaptation to the platform operable architecture. Google and the Open Handset Alliance make use of a virtual machine and language interpretation technologies. Their aim is to try to keep the GPL license domain of the integrated software

Open-Source Software Implications in the Competitive Mobile Platforms Market

119

packages outside the developer software development interfaces, most probably to avoid possible legal litigations. The Maemo 4 platform, in similitude with the Android platform, also integrates a large number of open-source components. However, it seems that Maemo architects are more satisfied with the original work from the open-source communities and therefore are not so intensively modifying the source code from integrated opensource projects. Moreover, Nokia allows its third party developers to directly access the integrated components programming interfaces. Thanks to the transparency of the Maemo open-source platform community, the following Table 3, illustrates how many packages are modified or directly integrated from the open-source projects. It is important to note that the number of packages are not correlated with the effort spent on its development, since each integrated project differs in size and complexity. Table 3. Core decisions taken by the Maemo architecture and integration teams Architecture Integration approach OSS directly integrated from communities (upstream) OSS modified by Nokia during the integration process OSS project initiated and developed under Nokia umbrella Closed source components developed under Nokia umbrella Closed source components by 3rd parties

Number of packages 68 79 49 92 2

Fig. 2. Venn diagram with open-source technology commonly used by the three platforms

120

S.Q. Mian, J. Teixeira, and E. Koskivaara

One of the most interesting findings of the research is a list of open-source technology components, which are commonly integrated by the vendor platforms studied. Illustrating which open-source components are commonly used by Apple, Google and Nokia the following Fig. 3 reveals high similitude between the Android and Maemo platforms. Open-source mature tools and libraries from the long established and reputed GNU and Apache communities are integrated across the three platforms. Google Android and Nokia Maemo integrate many common open-source components in their architectures, while Apple seeks a distinct architecture. Apple IOS and Google Android use the same open-source browsing technology webKit. Curiously, webKit is a fork of the open-source project KHTML imitated by Lars Knoll and George Staikos in 2006 within the Trolltech/Qt sphere, and that Trolltech was acquired by Nokia in the first semester of 2008. If vendors differ on what components they integrate, they differentiate even more on how they do it. Apple does not seem to be working “up-stream” with the communities of the open-source software. There are strong evidences that Google and Nokia work with and contract members from the open-source community. Both also try to integrate their own code modification/contributions to the source project. Forced by architecture decisions and dissatisfaction with the work produced by the open-source community, Apple and Google actively modify the packages they integrate. Nokia does it case by case, perhaps because Maemo architecture maps a more pure Linux-PC architecture from where most of the open-source software contributions come from. From the sharing perspective, the platforms completely differ on how their core blue-prints are shared as public domain with the open-source community. Apple seems to provide the source code of core components in its platform to avoid legal litigation. Nokia provides circa 80% of its Maemo platform, but hides components like hardware adaptation, network connectivity and UI elements like sounds and fonts. Google provides almost 100% of the platform source code, however, it delays the release of the source code for selected versions. This means that Google wants to have a momentum where it protects the blueprint of its latest developments, for example it delayed the release of the Android 3.0 Honeycomb (a tablet-oriented version) for few months, keeping its source code out of the public domain. It is important to note that the development does not follow the Bazaar model presented by Raymond (2001), both Nokia and Google work with the open-source community over the Internet but under very tight control. There are strong evidences that both vendors maintain the repositories with a public version of their platform and keep a closed internal one as well. Directly addressing the second research question, seeking a description of the open-source platform-based strategies employed by Apple, Google and Nokia, the following Table 4 captures key milestones employed by the vendors in concern. The presented information aims to provide a comprehensive detail of the different tactics in a platforms-war scenario that dominates the mobile-industry:

Open-Source Software Implications in the Competitive Mobile Platforms Market

121

Table 4. Open-source mobile-industry milestones 2005-2011 Year Apple

Nokia

Google

Maemo.org goes online Acquires Android Inc.

2005

2006 2007

Nokia 770 Internet Tablet Starts Shipping, OSS phone based on Maemo Buys YouTube Reinvents the Phone as iPhone Buys Navteq, the maker of digital mapping and navigational software Announced android, Founded Open Handset Alliance Achieves 40% phone market share Acquires Trolltech (owner QT technologies)

2008

Android was run on Nokia N810 Started Google I/O (Innovation in Open) annual conference Introduces the New iPhone 3G

Announces to make Symbian open source Announed the on-deck Open Content Distribution system where developers can sell their application direct to the users. World's first Android powered phone announced Made the entire Andriod source code, open source under the apache license; Launched G1 with HTC, first Android OS based mobile device

Apple releases iTunes version 8

Nokia acquires Symbian Ltd Introduces Google Latitude for mobile devices Samsung became the first among the top mobile manufacturers to adopt Android platform Ovi Store by Nokia is available globally to an estimated 50 million Nokia device owners Announces the New iPhone 3GS— The Fastest, Most Powerful iPhone Yet Releases android with HTC Hero device; Android Marketplace reaches 5000 applications Maemo 5 injects speed and power into mobile computing; Launches nokia.maemo.org

2009 Premieres iTunes version 9, New iPod touch, iPod shuffle and Ipod nano

Android Marketplace reaches 10000 applications

122

S.Q. Mian, J. Teixeira, and E. Koskivaara Table 4. (Continued) Accenture’s acquisition of Nokia’s Symbian Professional Services completed; Nokia announces official Qt port to Maemo 5; Sues apple for infringement of GSM, UMTS and WLAN standards Announces Over 100,000 Apps Now Available on the App Store Nokia releases Qt 4.6 Launches Ipad; App Store Downloads Top Three Billion

2010

Sues HTC for 20 patent infringements Sells Two Million iPads in Less Than 60 Days Releases iPhone 4; Sells Three Million iPads in 80 Days

Apple Introduces New iPod touch, iPod shuffle and Ipod nano.

Apple launches iOS 4.2 for iPad, iPhone & iPod touch; Sues Motorola over Multi touch in Android phones

Ovi Store delivers content and applications New Ovi Maps with free navigation races past 1 million downloads in a week; Merges software platform with Intel to form MeeGo Skype now available for Nokia smart-phones in Ovi Store Introduces N8 with OS Symbian^3 Sues Apple for infringement of 5 patents

Debuts ‘Touch and Type’ design with Nokia X3 phone Appoints Stephen Elop to President and CEO ; Started shipping N8 smart-phone Nokia started shipping C7 smartphone AMD joins MeeGo project

Buys AdMob, a mobile advertisement company Android Marketplace reaches 20000 applications Unveils Nexus One Phone

Android sales outpace iPhone

Skype becomes available for Android

Skype brings video calling to iPhone

2011

Apple’s App Store Downloads Top 10 Billion. Apple Launches Subscriptions on the App Store. Apple Launches iPad 2: Thinner, Lighter & Faster with FaceTime, Smart Covers & 10 Hour Battery; Launches iOS version 4.3

Larry Page replaces Eric Schmidt as CEO Adopts Windows Phone as primary smart-phone platform Sells QT commercial licensing business to Digia

Nokia's Ovi Store hits 5 million downloads per day; Announces plans to transfer Symbian software activities to Accenture

Debuts Firefox 4 for Android

Open-Source Software Implications in the Competitive Mobile Platforms Market

123

The IT industry is witnessing a war for the first time, which does not involve Microsoft and Nokia in a major role yet. When it comes to mobile device market, the rules of war have changed from the era which saw Microsoft dominating. Today’s mobile platform market is dominated by Apple and Google, with each of them employing a different strategy but more importantly resulting in success for them. Apple with their iPhone device and frequently releasing new versions has certainly got attention of the consumer interest. On the other hand, Google’s Android market share increased by almost 19% from 2009 to 2010, which is an indication of their strategic success (Gartner, 2011). Others players in the market such as Research in Motion undoubtedly control a fair market share but are now facing a new and distinctive type of competition which is more about innovation than surviving on an old success. In the very end, it is all about coming out with a new product which is very useful. A major aspect of strategy now resides on the open source, the contribution made by developers to promote innovation and richness in platforms and applications. According to Eric Schmidt, executive chairman Google, there are two kinds of players in the industry, the one who makes a very useful, focused product but is a closed competitor and that is Apple. Whereas Google is making technology available to everybody, sharing creativity, and making partnerships is taking a key step towards future innovation. Therefore, with more people involved, more investment will come, and eventually the consumer will choose the open competitor (Zakaria, 2011). However, Apple has been really successful in defining its products for focused groups, but they always try to own and control the working technology. After the success of iPhone 4, the iPad defined a new category of mobile devices. Apple tries to express its vision through its products and their success relies on the fact of not trying to do too much (Huang, 2011). A common success factor for both the companies is the use of open source software but they promote the concept at very different levels, considering the contribution they make in return. The great Nokia has already closed the Symbian foundation, while open for business it is not an open source anymore. They have also transferred the Symbian development operations to Accenture. A plus point for Symbian Qt developers is that they can port their applications to Android platform from now on, which benefits Google as well. Microsoft’s new partnership with Nokia, which makes their Windows 7 the primary platform and buying of Skype, makes their intention quite clear. But what strategy would they employ and whether it will be successful is yet to be seen. In order to address the third research question on how the third party developers are coping with the announced strategies, a compressive set of software development portals and forums were carefully analyzed according Romano et al. (2003) method for analysis of web-based qualitative data. Free form text communication from different software developers of mobile applications was collected, covering the first quarter of 2011 (January to March) for post classification and analysis. The following Table 5 presents the coding scheme used. A total of 4821 free form text sentences were coded using the dimensions “evaluation”, “intention to complement” and “desire for openness”.

124

S.Q. Mian, J. Teixeira, and E. Koskivaara Table 5. Coding scheme used as the reduction step described by Romano et al. (2003)

Evaluation Positive Negative Neutral Unknown

Intention to complement Will not develop for the platform Will develop for the platform Currently develops for the platform Will abandon platform development Neutral

Desire of openness Developer would like to access platform core source code Developer would like to be more included in the platform development Developer seeks a more open handset platform for final users. Developer seeks to run the platform in a different hardware, other than the one promoted by the vendor. Developer would like tighter control and more filtering efforts from the vendor.

From the set of collected sentences freely provided from software developers, a considerable number included a perceived evaluation statement from the developers towards the platform. A low number of these sentences, which either used many “jargons” used by software development communities or by being contradictory were coded in the “Unknown” category. It is clearly visible that the Android and Maemo platforms have been given positive remarks by the third party application developers. It is important to point out that many developers continuously provide positive or negative feedback reacting to the platform’s continuous development. The discussion is always fomented with the release of new developer tools and development interfaces from the vendors. Surprisingly, there is no evidence, that physical events targeting the software complementers communities such as the Apple Worldwide Developers Conference, the Google IO, the Symbian's developer's conference, and the Maemo Summit have immediate effects on the evaluation attitude of the platform software developers. The Fig. 3 below presents the positive and the negative reviews from the developers of different platforms. As visible, the most relevant finding is the low value perception from third party developers on the Symbian platform. Symbian is the only platform with higher negative reviews over positive reviews from developers. However, it is extremely important to notice that experienced developers provide more positive reviews on the Symbian platform as well. Many newcomers confront with the Symbian’s recent platform evolutions and leading technical capabilities in power and memory management. The inclusion of the Qt technologies in the Symbian platform captures most of the positive platform reviews. A really high number of the captured sentences were coded regarding the third party developers intentions in complementing the platform, especially on platforms that entered later in the market. The authors noticed that the developers often provide contradictory sentences regarding their “wishes” of complementing the platform; typically a developer provides rich discussion on its first contacts with the platforms. However after several weeks of “coding” a real world application, they change their initial intentions. This means that platform vendors can capture the attention of 3rd

Open-Source Software Implications in the Competitive Mobile Platforms Market

125

Fig. 3. Developers review by platform

party developers, but the platform value is only increased when the developer provides a valuable software package on top of the platform. As visible in Fig. 4 the intentions from Nokia in collecting contributions from third party developers to its Symbian platform seems to be unsuccessful. Many developers expressed their intentions in not adopting the platform and, even more concerning is the high number of developers stating that they will stop developing on the platform. However, it is surprising that many developers are willing to complement to the Maemo platform. Android seems to be the platform in position to collect high contributions from 3rd party software developers. In a market subjected to the forces of network effects, it will bring great value to the Android platform. Regarding the platform “openness”, a term referred to many times in the analysis on the Internet sites. It seems that Nokia sponsored platforms raise less debate over its platforms “openness”. The Apple iOS platform, on the other hand, is the one where developers more actively discuss its “openness”, curiously not from a software development point of view but from a critical end-user point of view. If open phones seem to be so important for critical end-users complementing the platform, it would be interesting to perform a future survey on what represents “openness” metric in the normal purchaser decisions. It is important to highlight that no iOS software developer revealed a desire in having the iOS platform running on devices not branded by Apple. All opinions expressed on “openness” revealed a desire towards a more openplatform, no software developer called for a tighter control and filtering in the iOS ecosystem. The Symbian developers also seem to be very satisfied with the hardware provided by Symbian phone makers such as Nokia and Samsung. As seen in Fig. 5

126

S.Q. Mian, J. Teixeira, and E. Koskivaara

Fig. 4. Developers commitment to each platform

Fig. 5. Developers perceived platform openness

none of the developers expressed a wish to run the platform on alternative hardware. Until the end of the first quarter of 2011, no Symbian developer had trouble in accessing the platform core source code. Perhaps the most interesting observation is the platform developers call for final-user empowerment. Heavy criticism was given on the impossibility of accessing, by legal means, with “root” privileges in handsets shipped with the iOS and Android platforms. Moreover, the fact that 3rd party applications can be only installed from the vendors Internet markets, commonly referred to as “app stores” or “app markets”, raises strong debate among platform software complementers.

Open-Source Software Implications in the Competitive Mobile Platforms Market

4

127

Limitations and Conclusion

Our study encompasses the generalized limitations inherent to case study research. From a set of single cases, the previous reported findings cannot be directly generalized to the current body of theoretical knowledge in Information Systems. However, as pointed by Yin (2002) and Flyvbjerg (2006), case studies play a central role in the academia via generalization as supplement or an alternative to other methods, even if “the force of an example” is still underestimated over formal generalization. This case study was a great opportunity for testing established knowledge in computer-based platforms from authors such as Anchordoguy (1989), Bresnahan and Greenstein (1999), Hagiu (2004), West (2003), and Gawer and Cusumano (2008). A heterogeneous and evolving definition of what computer-based platforms are; together with an increasing technical sophistication with increasingly complex layers; and an increased networked economy competition embracing phenomenon like openinnovation and co-competition; limit the generalization of the early knowledge built on top of the PC-industry to platform competition in the new mobile-industry. Even if it was not the main goal of this research, this case study can be useful for testing hypothesis gathered from previous literature. Future research should be performed on the concrete testing of some of the previously established research. It is quite obvious to notice high convergence between the PC and Mobile industries, where vendors try to explore the network effects associated with the compatibility between the different vendor portfolios. As an example, Apple like a traditional PC player is turning into a competitive mobile industry player, Nokia is already selling a lite PC with its netbook product and Google moved completely out of its core by providing a platform impacting the overall telecommunications industry. Vendors are explicitly managing customer lock-in and incompatibility strategies, as described by Shapiro and Varian (1999). One of the most notable example is provided by Ionescu (2009), where Apple continuously disables the access of competitor Palm pre device to its iTunes sync Internet service. On a curious note, Varian is now working as Chief Economist at Google. Our contributions provide a detailed description on how differently Apple, Goggle and Nokia make use of open-source software components and on how differently they cooperate with the communities of software developers with most of the credits in the integrated technologies. A detailed analytical view on the attitudes from community of software developers on the vendor platforms strategies provides academics and practitioners with valuable data to better understand the mobile-industry landscape.

References 1. Anchordoguy, M.: Computers Inc: Japan’s challenge to IBM. Published by Council on East Asian Studies, Harvard University (1989) 2. Bonaccorsi, A., Rossi, C.: Why Open Source software can succeed. Research Policy 32(7), 1243–1258 (2003) 3. Bresnahan, T.F., Greenstein, S.: Technological Competition and the Structure of the Computer Industry. The Journal of Industrial Economics 47(1), 1–40 (1999) 4. Flyvbjerg, B.: Five Misunderstandings About Case-Study Research. Qualitative Inquiry 12(2), 219–245 (2006)

128

S.Q. Mian, J. Teixeira, and E. Koskivaara

5. Gartner: Gartner Says Worldwide Mobile Device Sales to End Users Reached 1.6 Billion Units in 2010; Smartphone Sales Grew 72 Percent in 2010, http://www.gartner.com/it/page.jsp?id=1543014 (accessed May 27, 2011) 6. Gawer, A., Cusumano, G.: How Companies Become Platform Leaders. MIT Sloan Management Review 49(1), 28–36 (2008) 7. Hagiu, A.: A General Perspective on Multi-Sided Software Platforms and the Japanese Computer and Consumer Electronics Industries. Keizai Sangyo Journal - Research & Review (2005), http://www.rieti.go.jp/en/papers/ research-review/023.html (accessed April 1, 2011) 8. Hagiu, A.: Japan’s High-Technology Computer-Based Industries: Software Platforms Anyone?, Columns FY (149) (2004), http://www.rieti.go.jp/en/columns/a01_0149.html (accessed April 1, 2011) 9. Helander, N., Rönkkö, M., Puhakka, M.: Open Source Software Business in Finland Results of the National Survey. In: OpenMind 2008, Conference, Tampere, Finland (2008), http://www.sbl.tkk.fi/oskari/OpenMind2008.pdf 10. Huang, J.: Nvidia CEO: Why Android tablets aren’t selling | Nanotech - The Circuits Blog - CNET News (2011), http://news.cnet.com/8301-13924_3-2006294064.html?tag=mncol;posts (accessed May 27, 2011) 11. Ionescu, D.: Palm Lost the iTunes Sync Battle (Again). PCWorld (2009), http://www.pcworld.com/article/181057/ palm_lost_the_itunes_sync_battle_again.html (accessed May 26, 2011) 12. Lerner, J., Tirole, J.: The Economics of Technology Sharing: Open Source and Beyond. Journal of Economic Perspectives 19(2), 99–120 (2005) 13. Morris, C.R., Ferguson, C.H.: How architecture wins technology wars. In: Harvard Business Review on Managing High-Tech Industries, pp. 117–145. Harvard Business School Press (1999) 14. Ocock, T.: Guest post: Symbian OS – one of the most successful failures in tech history (2010), http://eu.techcrunch.com/2010/11/08/guest-post-symbianos-one-of-the-most-successful-failures-in-tech-history/ (accessed May 27, 2011) 15. Open Handset Alliance, 2011 – Alliance Overview | Open Handset Alliance 16. http://www.openhandsetalliance.com/index.html (accessed May 27, 2011) 17. Paajanen, A.: Open source software license decisions. Master Thesis in Information Systems Science. Helsinki School of Economics (2007), http://www.iteacosi.org/modules/docmanager/get_file.php?curent_file=10&cure nt_dir=2 18. Raymond, E.S.: The Cathedral and the Bazaar: Musings on Linux and Open Source by an Accidental Revolutionary. O’Reilly & Associates, Inc. (2001) 19. Romano Jr., N.C., et al.: A Methodology for Analyzing Web-Based Qualitative Data. J. Manage. Inf. Syst. 19(4), 213–246 (2003) 20. Seaman, C.: Qualitative methods in empirical studies of software engineering. IEEE Transactions on Software Engineering 25(4), 557–572 (1999) 21. Shapiro, C., Varian, H.R.: Information rules. Harvard Business School Press, Boston (1999) 22. Yin, R.: Case Study Research: Design and Methods. SAGE Publications (2002), http://www.amazon.ca/exec/obidos/ redirect?tag=citeulike09-20&path=ASIN/0761925538 23. Zakaria, F.: Eric Schmidt on Android vs iPad – Global Public Square - CNN.com Blogs (2011), http://globalpublicsquare.blogs.cnn.com/2011/05/19/ eric-schmidt-on-android-vs-ipad/ (accessed May 27, 2011)

Diminishing Print vs. Rising Electronic Publishing: The Case of Lithuania Gintare Gudaityte, Vida Melninkaite, Daiva Vitkute-Adzgauskiene, and Vladislav V. Fomin Vytautas Magnus University, Faculty of Informatics, Kaunas, Lithuania {g.gudaityte,v.melninkaite, d.vitkute-adzgauskiene,v.fomin}@if.vdu.lt

Abstract. In this work we undertake an interpretative case study to analyze the periodicals market in Lithuania. Specifically, we are exploring the possible reasons for the diminishing supply of and demand for traditional print press and whether or not electronic publishing can fill in the growing gap. In this work we use data collected through Internet based survey and interviews with publishing industry players to draw conclusions on the current and future trends in the epublishing domain in Lithuania. Keywords: publishing, electronic publishing, interpretative case study, consumer perspective, business perspective, e-Society, knowledge society.

1

Introduction

In 1999 Council of the European Union announced an ambitious vision of Europeans inhabiting Information Society by 2005 [5]. The vision of Information Society, dubbed as e-Europe or e-Society, was to bring forth “changes, the most significant since the Industrial Revolution” [5, p.2]. Looking at the history of national informatization processes in Europe and elsewhere one can notice that the task of building e-Society was far too often understood (or tackled) as bringing a large number of services online [6], i.e., seeing Information Society as a composite of the largest possible amount of online services. More than a decade down the road, we have learned the hard way that the success of establishing e-Society cannot be measured by the availability of (often very complex) services provided to citizens over the Internet. The sought for transformation of society requires distribution of knowledge among the citizens and businesses to eliminate knowledge asymmetry and facilitate market and democratic processes [4]. From the viewpoint of Information Society development, the publishing industry peresents a doubly interesting object to study. First, the media as it appeared in varying forms throughout the history has always been a source of information and knowledge to people. Second, the publishing services today are being moved to the electronic (online) formats and distribution channels. Thus, electronic publishing services become a cornerstone element of economy – information is at the same time a raw material and a product [4]. T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 129–137, 2011. © IFIP International Federation for Information Processing 2011

130

G. Gudaityte et al.

The extent of the transformation of traditional (printed) media to electronic one is most visible in developed countries, and the US in particular. In 2010 US residents were reading more news from online sources than from printed ones; the revenues from Internet advertisements overtook those from the print media. Mobile phones became one of the primary devices to bring news to the readers [11]. The new publishing paradigm is entering Lithuania, too. Few journals and newspapers started offering their content in specially formatted online editions; electronic reader devices are being introduced to the market. However, the transformation hasn’t acquired a momentum yet, the publishers are not rushing to set their foot into the new grounds – there is a lack of understanding of the business, technology, and legal aspects of the electronic publishing. There is a lack of confidence among the publishers that there is a demand for the new media, too. The advent of electronic media in Lithuania is taking place at the time when the demand for traditional press is diminishing. Despite having the largest number of mobile phones per capita in the world, Lithuanians lag behind many of their European neighbors in affordability and literacy when it comes to the use of e-services in general and electronic media in particular. Using interpretative case study, in this work we are trying to answer the question whether the diminishing offering of printed media is likely to be readily compensated by new electronic media services, thus preventing formation of “knowledge gap” on the way to building knowledge society in Lithuania.

2

Research Design

In this work we use an interpretative case study method. The aim of this research work is to provide an in-depth description and interpretative analysis of processes currently taking place in the publishing domain. The case study method is particularly suitable for the aims set, as it allows recording both the process-related data and at the same time interpreting the context in which the processes unfold [15]. Our data collection strategy is based on uses of multiple sources. Academic and popular literature, structured interviews, and Internet sourvey were used to obtain different perspectives on the phenomenon under investigation. Such process of establishing counterevidence is known as triangulation and refers to verification of research findings [2, 12]. Literature review was conducted to obtain an overview of the (global) ongoing transforation of the media business, capture the media-related trends in consumer behavior and technology and service offering. Internet survey was used to obtain deeper understanding of the perceptions of Lithuanian consumers with regard to the electronic media. Interviews with representatives of Lithuanian publishing business helped clarify the views of the media products’ and services’ providers on the emerging electronic publishing paradigm. The research was designed to cater for the following goals of the study: • Obtain the knowledge on (possible) problems/barriers associated with the introduction of electronic publishing services in Lithuania;

Diminishing Print vs. Rising Electronic Publishing: The Case of Lithuania

131

• Obtain the knowledge on the extent to which consumers accept electronic media (as an alternative to the printed one); • Be able to juxtapose data from different sources to analyze and interpret the perspectives of electronic publishing in Lithuania.

3

Case Study: Publishing in Lithuania

Electronic publishing has emerged as a global phenomenon. However, the lessons learned from the globalization of electronic commerce [7] suggest that (successful) business cases from e.g., US won’t necessarily work in another country. Economic, legislative, cultural, and demographic difference in the context of e-service’s development and/or application can render the successful service offering quite unuseful when transferred to another than the original context. Readers’s preferences and expectations and publishers’ innovation strategies play an important role in localization of global e-services. Based on our expert knowledge of the general state of affairs in the e-business domain in Lithuania, we set an initial hypothesis that Lithuanian population has not yet recognized electronic publishing as a primary type of media, and that Lithuanian publishers are not (technologically) ready to stimulate the demand by increasing the epublishing product offering. However, electronic publishing has a number of advantages over the traditional print media – hypertext, instantaneity, accessibility, among other. These advantages can stimulate the demand for the new type of media. In the following section we present a case study aimed to explore whether or not there are good prospects for the growth of demand on periodical e-publications in Lithuania. 3.1

Transitioning from Printed to Electronic Press

In 2009, there were 1285 periodical titles in Lithuania: 297 newspapers, 493 magazines, 96 bulletins, 382 other kind of periodicals [9]. In 2010 the list got shorter by 28 titles. Newspapers saw the biggest impact, loosing 9% (44) of its 2009 titles or 23% (53,8 thousand) in printed coppies per annum [9]. At the same time the number of residents in Lithuania during one year decreased only by 1,6% [13], making the relative decrease in the number of newspaper titles per capita 16. Journals titles’ list got shorter by 5, whereas the print volume decreased by 12%, or 4 journals fewer per capita adjusted to the decrease in population [8]. These numbers show the downward trend in traditional (printed) media offering. Looking at the electronic media market, there is a reverse trend – the number of registered titles has increased by 19 since 2008 to reach 259 [13]. There is also a limited number of titles, which are issued in both electronic and printed format. Three out of four most visited websites were exactly those “mixed” titles: Lrytas.lt (“Lietuvos rytas” newspaper) was #2 receiving 42% of daily Internet users, while Balsas.lt (“Balsas” newspaper) was #3 with 40%. #4 was 15min.lt (“15 min” free newspaper), which was receiving 39% of the Internet users. The most visited website – Delfi.lt – electronic-only news portal – was receiving 55% of all Internet users in the beginning of 2011 [8].

132

G. Gudaityte et al.

Looking at the variety of printed press, in 2009 the most popular categories were tourism and sports (139), women and families (105), professional (76), political, philosophy, and religious (64), popular science (58), business (41), and children and youth (42). The most popular single periodical title was “TV Antena“, which is printed as a supplement to the Saturday’s issue of “Lietuvos rytas“ newspaper. The runnerafter was a “yellow press” magazine “Žmonės“ (Lith.: “People”). The third place was held by the newspaper “Lietuvos rytas“[14]. Unlike countries like the US [11], electronic publishing in Lithuania today is a niche market. However, there are obvious signs of change – the traditional print, radio and TV media is extending its presence in the cyberspace, often through a lowerquality and free-access offerings, aiming at reaching out to a larger consumer population. The free-access offering of the e-publishing pioneers is contributing to the formation of such consumer attitude that the freely available on the Internet e-media is dubbing the information presented in the pay-per-view or pay-for-subscription press. This means that the pioneers of e-publishing are cannibalizing on their own printed press market (if such existis) or setting consumer expectations to have free access services forever. This happened to the popular Lietuvos rytas newspaper – since the newspaper started offering a free electronic verion of the newspaper, the sales volumes for the printed newspaper fell 5%. Despite the seeming optimism which which Lithuanian publishers embraced the opportunities brought about by e-publishing, we are tempted to assume that the freeaccess e-publishing business can be sustained in Lithuania neither in the long, nor in the short run. These assumptions are based on the analysis of the current market trednds in Lithuania, as follows. Since the 2008 financial crisis, revenues the publishers were earning from advertisements have fallen drastically, subscriptions to periodicals are in decline. Several traditional journals and newspapers which started offering about or almost the same content online as in their printed press, have seen decrease in subscriptions and sales of the printed issues. A peculiar trend has emerged – the higher quality outlets, in order to keep up against the declining revenues, started offering lower quality news, which drew their traditional customers dissatisfied and eventually abandoning the outled. On the other hand, the media outlets which initially positioned themself as free-access and low quality news service see a growing demand for their offering [1]. While the service offering of electronic press market in Lithuania is very limited, the observed trends suggest there is a potential for growth. Recognizing the importance of e-services in stimulating knowledge-based economy, the state is supporting the move towards e-services with such state-funded infrastructural investment projects as e.g., the one aimed at development of innovative e-reader software and the application of it [3]. The project lasted for six months during which technical and market aspects of electronic publishing were studied. One of the conclusions the project came to was that electronic readers and paid access electronic media will become more popular when the price the consumers have to pay for the device will fall undet 250 Litas (ca. $100). This thesis is partially supported by evidence from UK, where recent market research, quite contrary to the results of the early e-commerce

Diminishing Print vs. Rising Electronic Publishing: The Case of Lithuania

133

days [7], showed the elder, not the younger part of the population are the early adopters of e-services and e-reader devices [10]. So, 6% of people in the age of over 55 were reported to be using e-readers, while only 5% in the age group of 18-24 [10]. 3.2

Survey Data Analisys

The survey was conducted using an online Internet survey management tool. An invitation to participate in the survey and the actual web link (URL) to the survey were placed on the Lithuania e-survey portal apklausa.lt, as well as distributed by email to various groups and individuals. Access to the survey was open to anyone having the URL, thus making it possible to create a snowball effect in promoting the survey – any one recipient of email with the URL for the survey could forward it further to anyone else. The survey was open between March and May 2011 and generated 562 responses. The survey comprised of 18 fixed-response and open-ended questions, aimed at gathering information on citizens’ preferences with regard to print and electronic media (services) consumption. Of the total 562 respondents, 31% were male, and 69% were female. The largest age groups were 18-25 (60%), 26-35 (22%), and 36-50 (11%). 50% of the respondents had a university degree, 17% – special professional, and 17% – secondary education. 40% reported to have monthly income below 1000 Litas ($400), 17% – 1000-1500 Litas, 13% –1500-2000 Litas, and 16% – more than 2000 Litas ($800). The first section of the survey dealt with the question of the frequency of and the purpose for the use of Internet. 81% of respondents are using Internet on the daily basis, while 13% – 5-6 days a week. 92% of the respondents use the Internet to search for information, 80% – for online banking, 68% – to read newspapers and online magazines, or to download respective content. 69% use the Internet for their studies or education, while 58% – for their work. 41% use the Internet to listen to radio, watch TV. 60% – to watch/download movies or play/download games. Answers to the questions on buying frequency and monthly budget for print media are presented in (Table 1). Table 1. Frequency of buying the printed media and monthly spending for it Monthly spending on print media, Lt Buying frequency

Always Sometimes

0 1

10 14

10-20 35

20-30 13

30+ 20

Total: 82

0

203

88

15

10

316

Never

164

0

0

0

0

164

Total:

165

217

123

28

30

562

134

G. Gudaityte et al.

The results show that the majority of respondents are sometimes buying printed media products, not spending on that more than 10 Litas ($4) per month. The second largest group is sometimes buying for the amount of 10 to 20 Litas ($4-$8) a month. While only 30 respondents (5%) are ready to spend more than 30 Lt ($12) per month, and not more than 35 (6%) report to always buy a print media, irrespective of their budget for it. Answering the question, whether traditional (print) or electronic form of publications are given preference, there are about twice as many respondents giving preference to electronic media in the education categories up to the secondary education (inclusive), while for the university education the rate is more than four to 1. Overall, only 125 respondents (22%) prefer printed media to electronic one. When asked whether they feel a need to use electronic publishing products, 73% of respondents answered “yes”. When looking at the frequency and duration of reading from electronic sources, the distribution of responses appears to be quite even across the categories (see Table 2). Table 2. Frequency and duration of reading of electronic media sources Average time of reading (for each access event)

Frequency of reading

<15 min

15-30 min

30-60 min

>60 min

0 min

Few times a day

11

55

65

18

0

Total: 149

Once a day

10

85

54

6

0

155

Weekdays only

3

17

16

3

0

39

2-3 times a week don’t read

47

67

93

12

0

219

0

0

0

0

0

0

Total:

71

234

218

39

0

562

At least once a day electronic media is reported to be read by 304 (54%) respondents, most of them reading for 15 to 60 min each time. Another section of the survey deals with the types of electronic periodicals. 27% of respondents read online/electronic magazines and 25% – newspapers or news. Other popular types of media are dictionaries, vocabularies, encyclopedia (20% of respondents). Only 1% of respondents reported not to be interested in any kind of elelctronic media. To find out what qualities of electronic media are the most valued by the reader, questions were asking on the content of electronic publications. Among the most important characteristics for electronic media respondents name trustworthyness of information (74%) and (relevant) content (71%). The following in the importance characteristic is the non-sophisticated presentation of the information (39%). This includes easy access and intuitive layout of the electronic publication. Interactivity is ranked almost evenly as important (39%) and not so important (35%). This may suggest that the readers are not used to (highly) interactive design, preferring simplicity of presentation to that of reachness of features.

Diminishing Print vs. Rising Electronic Publishing: The Case of Lithuania

135

Peculiarly, for the majority of respondents, violation of copyright is not an important issue – only 40% rated this factor as important or very important, while 26% – as non-important, or totally unimportant. The price of electronic press in Lithuania is currently set at about the same level as for similar printed editions. Therefore, it was asked in the survey whether respondents would be willing to buy electronic media if its price was considerably lower than that of the print press. The answers received show that Lithuania readers aren’t ready to pay for electronic media – 45% of respondents aren’t sure they need electronic publications at all. 29% would have bought electronic publications if these were cheaper than the printed ones, while the choice being constrained by multiple other factors, such as the publication type/subject, its accessibility and the price. Another important question relates to the availability and popularity of e-reader devices. 78% of respondents have heard about (the existence) of e-reader deveices, while 17% - haven’t. Only 5% reported to have one. When asked, what could stimulate their interest towards electronic press, the majority wished to have a wider choice of electronic media (27%) and lower prices for it (26%). 23% wished to have a wider range of and a lower price for e-reader devices, while 11% wished the digital rights management and copy-protection systems weren’t as stringent. When asked whether they see the electronic press as having advantage over the printed one, 28% gave positive answers and 27% – negative. The remaining 45% saw the two forms of publishing as equal. 3.3

Summary on the Data Analysis

The survey data analysis shows that the Internet usage in Lithuania is high, people use the Internet to search for information, use e-banking services, read newspapers and magazines. There is a diminishing demand for traditional print media, as users are not willing to spend more than 10 Litas ($4) per month on printed periodicals. At the same time, the need for electronic media is high. Most read categotries of electronic periodicals are magazines and newspapers. For reading electronic media, people allocate a substantial portion of their time on a daily basis. When choosing what to read (online), respondents give preference to the quality and content of information source. Whether or not the information is obtained lawfully is of little concern to Lithuanian reader. The use of e-readers is deminitive, although the majority of the readers are aware that such devices exist. Lower price for e-reader, lower access fee for electronic media, and wider choice of electronic publications – these are named as factors that will stimulate the growth of e-publishing services. At the same time, respondents evaluate the importance of electronic and print press as equal.

4

Conclusions

Having conducted case study and data analysis, we can conclude that Lithuania, similar to many other countries, currently transitions from printed press market to electronic one. On the backdrop of decreasing demand for traditional media, Lithuanian

136

G. Gudaityte et al.

publishers are likely in the short run to face two alternatives – shift to electronic press or close their business all together. The concept of electronic publishing is often compared to that of the traditional one. Features and business models, that are typical to traditional media pertain to electronic media, too. Therefore, many drivers and barriers which can be found in traditional press business are relavant for the electronic press. In one possible setup, electronic periodical can be simply mirroring a printed publication, where the content of the printed document, page layout, etc. are replicated. In another setup, electronic periodical can exist without its printed twin. The interview data analysis allowed us to assume that publishers in Lithuania are sceptical about the business prospects of electronic publishing. At the same time, they understand that new technologies are advancing rapidly and changes in the publishing domain are inevitable. Small market size, plagiarism, lack of legal basis, lack of initiatives and creativity are named as the biggest barriers for publishers in deciding to transition to electronic press. A conclusion can be drawn, that the main issue for owners of electronic publishing in Lithuania is the lack of (knowledge on) sustainable business strategies and transition models. Also, the supply of e-reader devices is currently too low, and their price is too high. Additional technical issue to consider is the lack of knowledge on feasible solutions for content safety and format compatibility. Our conclusions are supported by the fact, that the absolute majority of electronic versions of Lithuanian periodicals is implemented in a form of an online hypertext twin of the printed version. Such setup is limiting business possibilities of the publishers due to commonly reported unwillingness of Lithuanian internet users to pay for information access services. Commercial ads in this case won’t lead to a sustainable business model due to overly small market segments for most periodicals.

5

Managerial Implications

Finalizing, we can draw few managerial implications for the e-publishing business: • The uptake of e-publishing in Lithuania can be stimulated by introducing e-reader devices and developing e-reader versions of periodicals. Use of e-reader for distributing electronic versions of periodicals would allow to use business models similar to the ones used for traditional print media, thus reducing the economic risk in the context of overly small publishing markets. • The use of e-readers for the distribution of electronic press would limit the uncertainties the publishers are facing with regard to unauthorized (re-)distribution of media by users/customers. The hardware-based media distribution approach would provide the copyright owners with better possibilities and tools for copy-, redistribution- and access control. • Building on the experience of successful business model of mobile phone subsidization by mobile operators, publishers of electronic media could offer a possibility to purchase e-reader devices at a lower price, given that consumers subscribe to receive paid electronic publications for a certain period of time. • Traditional press publishers are encouraged to publish on the Internet, at the same time, as electronic publications serve as advertisement for the printed one.

Diminishing Print vs. Rising Electronic Publishing: The Case of Lithuania

137

Presence in the electronic domain also creates possibilities for establishing feedback sites with readers. However, e-reader version should be considered in addition to the online version. • Owners of existing and future electronic periodicals should be very careful in selecting (publication) standards in order to attract larger audience at the backdrop of the existing diversity of e-reader devices. Following common European and Lithuanian information infrastructure development frameworks, we recommend to use .pdf and .epub formats for electronic periodicals, as these are based on open standards and thus can support larger variety of devices. • The use of open standards is recommended, their advantage being in the ease of access of corresponding electronic publications. This results in the increase in freedom of choice for the end user, avoiding the promotion of one single producer.

References 1. 15min.lt, Trečdaliu auga, 15 min tiražas (The circulation of 15 min grows by one third). 15. 2011, 15min.lt: Kaunas 2. Adams, G.R., Schvaneveldt, J.D.: Understanding Research Methods, 2nd edn. Longman Publishing Group, N.Y (1991) 3. Agmis.lt, Agmis pasirašė sutartį dėl ES iš dalies finansuojamų elektroninio formato leidinių mokslinių tyrimų (Agmis signed agreement on scientif research reagrding partial ES financing for electronic format publications), Vilnius (2010) 4. Castells, M.: The Rise of the Network Society, 2nd edn. (2000); The Information Age: Economy, Society and Culture, vol. I. Blackwell Publishers, Ltd., Oxford (1996) 5. Council of the European Union, eEurope. An Information Society for All (1999) 6. Daniel, E.M., Wilson, H.N.: Dynamic capabilities in e-business transformation. European Journal of Information Systems 12, 282–296 (2003) 7. Fomin, V.V., King, J.L., Lyytinen, K., McGann, S.: Globalization and E-Commerce VII: Environment and Policy in the U.S. Communications of the Association for Information Systems (CAIS) 10(Article 8), 276–325 (2003) 8. gemiusAudience, gemiusAudience: The most reliable source of knowledge about the Internet and its users (2010), http://www.audience.lt 9. Lietuvos Nacionaline Biblioteka, Lietuvos spaudos statistika 2009 m. (Lithuanian National Library, Statistics of Lithuanian Press in 2009), Lietuvos nacionalinė Martyno Mažvydo biblioteka (Lithuanian National Martynas Mazvydas Library) (2010) 10. E-book popularity set to increase this year (2011), Redsauce.com 11. Rosenstiel, T., Mitchell, A.: The state of the New Media 2011. An annual Report on American Journalism (2011), http://stateofthemedia.org/ 12. Stake, R.E.: Case studies. In: Denzin, N.K., Lincoln, Y.S. (eds.) Handbook of Qualitative Research, pp. 236–247. Sage Publications, Thousand Oaks (1994) 13. Statistics Lithuania, Statistical Yearbook of Lithuania 2010, Statistikos departamentas (Department of Statistics) (2010) 14. TNS.lt, Baltijos šalių leidinių skaitomumas 2010 ruduo (Readability of Baltic country publications, Autumn 2010), Vilnius (2011) 15. Yin, R.K.: Case Study Research. In: Design and Methods, 2nd edn. Sage Publications, Inc., Thousand Oaks (1994)

A Method for Conceptual Modeling of Semantically Integrated Use-Case Scenarios Remigijus Gustas and Prima Gustiene Department of Information Systems, Karlstad University, Sweden {Remigijus.Gustas,Prima.Gustiene}@kau.se

Abstract. A use-case is specified as a set of possible scenarios of interactions. Scenarios can be decomposed into workflows on different granularity levels. Use-cases are fundamentally a text-based documentation form written in natural language text. The textual description of a complicated scenario can be ambiguous, incomplete and inconsistent. In this paper we demonstrate a conceptual modeling method for representing use-case descriptions by using a graphical language. Simple interaction loops are viewed as fundamental elements for composition of scenarios. Each interaction loop is analyzed separately and therefore it provides a natural way of decomposition. Modeling of overlaying interaction flows between organizational and technical components enables separation of crosscutting concerns in system engineering without requirement to specify a complete solution. The goal of this paper is to demonstrate the advantages of conceptual modeling approach, which allows to introduce evolutionary extensions and to construct use-case scenarios with a comprehensible structure. Keywords: Interaction dependencies, separation of concerns, basic pattern of a transaction, interaction loops, scenarios.

1

Introduction

Use-cases are a way to capture requirements. It is the form of requirements engineering [1]. According to Cockburn [2] a use-case is a description of the possible sequences of interactions between the system and its external actors. Each sequence of interaction events can be viewed as a scenario. Every use-case contains a set of possible scenarios related to the goal of a primary actor. Historically use-cases descriptions are written as text-based documentation. Natural language is verbose and flexible to use in conversations, but it is really problematic when it comes to system modeling as natural language is notoriously ambiguous in its meaning [3]. Especially, complex scenarios may result in ambiguous, incomplete and inconsistent textual descriptions. According to Jacobson [4], use-cases can be seen as different stakeholders concerns, which are important to understand while exploring and collecting the requirements for the system. Very often a certain concern is spread across multiple components. It means that the realization of functional requirements, which are usually specified as use-cases can cut across multiple components. Inability to keep crosscutting concerns separate causes problems for system designers when the requirements are modified. T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 138–152, 2011. © IFIP International Federation for Information Processing 2011

A Method for Conceptual Modeling of Semantically Integrated Use-Case Scenarios

139

The designer must identify all related components and to find out how these components are affected by introduced changes. Especially, modifying the requirements, which are related to a big number of diagrams, is quite problematic. Poor understanding of crosscutting concerns makes it difficult to make even simple evolutionary extensions of information system (IS) specifications. The scope of scenario can vary. It may include all business events, or it may include just some events, which are of interest to one specific actor. Scenarios can be used to define workflows on different granularity levels. Our studies indicate that simple interaction loops [5] can be viewed as fundamental elements for the composition of scenarios. Each interaction loop can be analyzed separately as it is required by the principle of separation of concerns. In such a way, interaction loops provide a natural way of decomposition. Two related loops can be used for analyzing integrity between static and dynamic aspects of some scenario. They are also useful as a concern composition mechanism. A scenario is an excellent means for describing the order of interactions. It can be conceptualized as a combination of several loops between a set of loosely coupled actors. Use-cases can be described on different levels of abstraction and they can be combined into scenarios in various ways. In the object-oriented modeling approaches, a scenario is typically specified by a textual narrative description. A textual description of a complicated scenario can be incomplete and inconsistent with other representations. One of the goals of this paper is to demonstrate how use-case scenarios can be expressed by using Semantically Integrated Conceptual Modeling (SICM) method [6]. This approach puts into a foreground the modeling of interactions [7] among actors [8]. Interaction dependencies are used to preserve the modularity of concerns and to integrate behavioral effects with structural changes in various classes of objects. Use-cases can be viewed as slices, which are analogous to overlays. Such overlays can be stacked on top of each other. In this way, overlays can be understood as a concern separation technique [4]. Use-case slices define the behavior that extends the element structure. So, it looks like use-case composition mechanism and concern separation principles are quite obvious. However, introducing evolutionary changes, which are related to a big number of diagrams, is quite problematic in object-oriented approach using Unified Modeling Language (UML). It is common to all system analysis and design methods to separate disparate views [9], because a human mind allows focusing on a particular type of diagram at a time. Designers are typically dealing with one type of diagram, which defines behavioral, interactive or structural aspects of a system in isolation. Therefore, it is difficult to take into account semantic dependencies between the static and dynamic aspects related to a particular concern in a very early modeling phase. That is why most conventional IS design methods are not so useful for the detection of inconsistency or incompleteness in various use-cases specifications. In this paper, we will demonstrate how interaction dependencies can be used to construct unambiguous graphical descriptions of scenarios with sequential, iterative, synchronized and alternative behavior. Most of information system methodologies are quite weak in representing the alternative scenarios and the consequences if commitments between actors are broken.

140

2

R. Gustas and P. Gustiene

Use-Case Descriptions

Use-case diagrams can be produced in two steps. The first task of the system analyst or project team member is writing use-case descriptions by using a natural language text. The second task is to translate the use-case descriptions into use-case diagrams. A use-case description should contain all needed information for building other UML diagrams. Unambiguity of use-case descriptions is important to validate the semantic integrity of the diagrams. The problem is that the narrative text, which defines flows of events of different use-cases, can be ambiguous, incomplete and inconsistent. There are three types of event flows, which are documented for a use-case: 1) Normal flow of events, 2) Subflows, and 3) Alternate flows. A use-case can be understood as a transaction. Any transaction can be analyzed as a simple workflow loop [5], which captures service value exchange between two or more parties. Both requests and responses are viewed as necessary business events. According to Ferrario and Guarino [10], service interactions are not objects or products, they are events. Service responses cannot be delivered without initiating service requests. A response can be viewed in a number of ways. It can be represented by a promise to deliver a desirable result to service requester or it can be viewed as statement, which brings a desired value flow [11] to service requester. Any workflow loop indicates that service provider receives service request and transforms it into service response. Service requester, request, service provider and response are minimal set of elements for defining any type of a service interaction loop. It is illustrated graphically in figure 1.

Fig. 1. Basic interaction loop

The presented elementary interaction loop is a basic element, which can be used in a very early conceptual modeling phase for separation of concerns in information system analysis and design. Two loosely coupled actors will be represented by the following expression [12]: If Request(Service Requester Service Provider) then Response(Service Provider  Service Requester). We will demonstrate the conceptualization of normal flow of events and alternate flows for the slightly modified case study example, which was analyzed by Jacobson and Ng [4]. In this example, the Reserve Room use-case is extended by the Handle Waiting List use-case. Both use-cases are represented in figure 2.

A Method for Conceptual Modeling of Semantically Integrated Use-Case Scenarios

141

Fig. 2. Example of use-case diagram

The Reserve Room use-case scenario can be graphically defined by two interaction loops between Customer and Hotel Reservation System. The primary interaction loop is composed of the underlying interaction loop. Both loops of the Reserve Room usecase scenario are graphically defined in figure 3.

Fig. 3. Elementary interaction loops of the Reserve Room use-case

The presented service requesters and service providers are actors. Actors are active concepts, which can only be motivated by a set of interaction dependencies that keep the enterprise system viable. Interaction dependency R(AB) indicates that actor A is an agent, which can perform action R on one or more recipients B. Interaction dependencies in the diagrams are graphically indicated by broken arrows. Broken arrows denote moving flows between actors such as information, decisions or materials. Actors are represented by square rectangles and actions are represented by ellipses. All actions are used to show the legal ways in which actors interact with each other. The first interaction loop (it is graphically represented in figure 3,a) cannot be executed without triggering the underlying interaction loop on the more specific granularity level (figure 3,b). The second interaction loop is initiated by the Hotel Reservation System. Use-case scenarios are conceptualized by identifying the flow of interaction events. Each event makes about the same amount of progress towards completion of a usecase. The equivalent narrative text of the Reserve Room use-case normal flow of events (see figure 3) can be described as follows: 1) The Customer requests room by entering the specific room requirements (including the desired period of stay), 2) The Hotel Reservation System offers various available rooms with different rates,

142

R. Gustas and P. Gustiene

3) The Customer selects the available room and enters the required information about an expected room guest. This step can be repeated several times. It means that more than one room can be selected, 4) The Hotel Reservation System creates reservation with the details of all selected hotel rooms, displays reservation information to the Customer and consumes the selected types of available room. Both graphical and textual descriptions in essence define the Reserve Room use-case as the flow of interaction events. Events are identified by using two kinds of guidelines as far as the syntactic and semantic structure of sentence is concerned. The syntactic guideline has something to do with the form of sentence. Each individual event should be expressed as subject-verb-object and, optionally, preposition-indirect object [13]. Such constructions are useful in identifying actors, operations and classes. The second set of guidelines is related to the semantic roles various concepts play in the sentences. Normally, each event should define the requester or performer of the action. The use of request–response type of sentences results in user-centered requirements documentation that can be useful for user training and testing. If the use-case description is too complex, it should be decomposed into a set of more simple flows. Subflows or alternate flows can be used for this purpose.

3

The Basic Constructs of SICM Approach

Event flows can be defined in terms of essential interactions between organizational or technical components. Technical components correspond to enterprise subsystems such as machines, software and hardware. Organizational components can be humans, organizations, their divisions or roles, which denote the groups of people. Interaction dependencies among actors are important for the separation of crosscutting concerns. By following interaction dependencies, it is possible to explore various ways in which enterprise system components are used. Event flows can be analyzed as a set of workflow loops [5]. A workflow loop in SICM method [14] is considered as a basic element of scenario, which describes interplay between service requesters and service providers. In its simplest form, any workflow loop is viewed as a response to request that creates promise or provides a value to service requester. Interaction dependencies are extensively used in a foreground of enterprise engineering methods [7]. These methods are rooted in the interaction pattern analysis and philosophy of language. The purpose of introducing them was initially motivated by the idea of creation computer-based tools for conducting conversations. The goal of this paper is different. We are going to demonstrate how to apply the interaction dependencies in combination with the set of semantic dependencies, which can be used for the graphical description of use-case scenarios. The sequences of interaction events are crucial for analyzing scenarios, which are expressed in terms of requests and responses between actors. For example, Create Reservation action can be viewed as a promise in connection to Request Room action.

A Method for Conceptual Modeling of Semantically Integrated Use-Case Scenarios

143

Behavioral and structural aspects of interactions can be analyzed in terms of their reclassification, creation or termination effects. When two subsystems interact one may affect the state of each other [15]. Structural changes of objects are defined in terms of object properties [16]. Interaction dependency R(AB) between two active concepts A and B indicates that A subsystem can perform action R on one or more B subsystems. An action typically manipulates properties of objects. Otherwise, this action is not purposeful. Property changes may trigger object transitions from one class to another. The behavioral effects of communication actions are expressed by using transition links (──) between various classes of objects. Reclassification of object can be defined in terms of communication action that is terminating an object in one class and creating it in another class. Sometimes, objects may pass several classes, and then they are terminated. Graphical notation of the reclassification construct is graphically represented in figure 4.

Fig. 4. Graphical notation of reclassification

Unbroken arrows indicate control flow of creation and termination effects. Object classes represent a persistent or transient set of objects. Fundamentally two kinds of changes are possible during any reclassification: termination and creation of an object. A creation is denoted by outgoing transition arrow to a post-condition class. A termination action is represented by a transition dependency directed from a precondition object class. Before an object can be terminated, it must be created. A precondition class in the termination construct is understood as final. For instance, when the Hotel Reservation Request is created, it can be reclassified to the Hotel Reservation by using Create Reservation action. Structural changes of objects are manifested via static and dynamic properties. Dynamic properties are represented as actions, which are connected to classes by creation and termination links. Static properties can be represented by the mandatory attributes. The mandatory attributes are linked to classes by the single-valued or by multi-valued attribute dependencies. One significant difference of the presented modeling approach is that the association ends of static relations are nameless. The justification of this way of modeling can be found in some other papers [12], [17]. Semantics of static dependencies are defined by cardinalities, which represent a minimum and maximum number of objects in one class (B) that can be associated with the objects in another class (A). Single-valued dependency is defined by the following cardinalities: (0,1;1,1), (0,*;1,1) and (1,1;1,1). Multi-valued dependency denotes either (0,1;1,*) or (1,1;1,*) cardinality. Graphical notation of various static dependencies is represented in figure 5.

144

R. Gustas and P. Gustiene

Fig. 5. Notation of static dependencies between concepts

The diagram presented in figure 3 does not provide any semantic details of control flows between communication actions. It shows only the necessary events in a Room Reservation scenario. The actions such as Request Room and Create Reservation should also specify the acceptable ways for structural changes to occur in different classes of objects. In general, communication actions can be sequential, iterative, alternative or synchronized with the secondary workflow loops. Triggering conditions of the secondary interaction loops may depend on the objects, which are created or terminated in the overlaying interaction loops. Pre-condition and post-condition classes are crucial to understand the dynamic aspects of interactions. A created object in one loop can be consumed in another. The creation and termination of objects allows constructing scenarios, which are enclosing optional or mandatory workflows. Overlapping classes can be used to synchronize interaction loops together. For instance, Create Reservation action terminates Hotel Reservation Request object, which was created as a result of Request Room action, and creates Hotel Reservation object. These creation and termination effects are graphically described in figure 6.

Fig. 6. Overlaying description of the Reserve Room use-case scenario

The corresponding Hotel Reservation and Hotel Reservation Request objects are composed of parts, which must be terminated and created at the same time. In this way, creation and termination effects define constraints on various types of objects in sending and receiving interaction flows between actors. Inheritance, composition and mandatory attribute dependencies can be used for reasoning about the consequences of object creation and termination effects. According to the conceptual modeling rules [6], the creation of Hotel Reservation is causing creation of at least one Hotel

A Method for Conceptual Modeling of Semantically Integrated Use-Case Scenarios

145

Room[Reserved] and expected Guest. It should be noted that the semantic power of UML object flow and sequence diagrams combined together is not sufficient for capturing the equivalent effects.

4

Composition of Interaction Loops

Elementary interaction loops can be viewed as fundamental elements for defining scenarios [17]. Graphical representation of scenario is an excellent means for describing the order of interactions. More specific underlying interaction loops can be analyzed in the context of the overlaying loops on higher granularity levels as it is required by the principle of separation of concerns. It means that the scope of scenario can vary. A scenario may include just some events, which are of the interest to one specific actor. However, it can naturally be linked to all other overlaying business events. The natural language descriptions of such complex use-case scenarios require a lot of supplementary information, which specifies pre-conditions, post-conditions and other special constrains for inserting subflows into main flow of events. In this section, we demonstrate how to construct the graphical descriptions of scenarios with a more comprehensible structure. The Reserve Room use-case scenario can be graphically defined by a number of interaction loops between Customer and Hotel Reservation System. The primary interaction loop (see figure 6) can be viewed as an overlay of secondary loop on more specific granularity level. Composition of two interaction loops of Reserve Room usecase scenario is presented in figure 7.

Fig. 7. Integrated conceptual representation of the Reserve Room scenario

146

R. Gustas and P. Gustiene

The primary loop of the Reserve Room use-case is synchronized with the underlying interaction loop. It defines the functionality of offering and selecting available hotel rooms. The underlying interaction loop is as follows: If Offer Rooms(Hotel Reservation System Customer) then Select(Customer  Hotel Reservation System). According to the presented control flow, the Select room action cannot be triggered prior to Offer Rooms action. Select action can only be performed in parallel with the Create Reservation action, because the creation of Hotel Reservation is synchronized with the creation of its compositional part Hotel Room[Reserved]. The termination of Hotel Reservation Request is synchronized with the reclassification of Type of Room[Desirable] to Offer, which is composed of at least one Hotel Room [Available]. A simple interaction loop between service requester and provider in the SICM method is viewed as the basic element of any communication process [5]. Interaction loops may be composed together into more complex interaction webs by using creation and termination links. If the object transition effects cannot be conceptualized by using pre-condition or post-condition classes, then the communication action is not purposeful. Interaction dependencies without purposeful actions make no sense and should be eliminated. The modeling of interactions and object transition effects together is critical for reaching semantic integrity among static and dynamic aspects of IS specifications. In the presented example, object creation, termination and reclassification effects show very important semantic details of unambiguous scenario in which two interaction loops are composed together. The more specific workflow loop is underlying the primary interaction loop. The underlying loop is required for the selection of desirable room type and for providing necessary data about guest. Underlying loops can be mandatory or optional. By following the interaction dependencies between actors, designers are able to understand the creation and termination effect in various classes of objects. In this way, the transition links are used to capture the dynamic dependencies between interaction loops.

5

Bridging from SICM Constructs to the Basic Pattern of a Transaction

Interaction dependencies are successfully used in the area of enterprise engineering [7]. Our intention is to apply the interaction dependencies in combination with the conventional semantic relations, which are used in the area of system analysis and design. Interaction loops can be expressed by interplay of coordination or production events, which appear to occur in a particular pattern. The idea behind a conversation for action schema [18] can be explained as turn-taking. Requester (R) initiates a request (R: Request) action and then is waiting for a particular promise (P: Promise) or a service provision (P: State) action from Performer (P). Request, Promise and Acceptance are typical coordination actions, which are triggered by the corresponding types of basic events. Coordination events are always related to some specific production

A Method for Conceptual Modeling of Semantically Integrated Use-Case Scenarios

147

event. Both coordination and production events can be combined together into scenarios, which represent an expected sequence of interactions between requester and performer. We will show how creation, termination or reclassification constructs of the SICM method can be used to define the new facts, which result from the main types of events of the basic transaction pattern [7]. Four basic events and related reclassification effects are represented in figure 8.

Fig. 8. The basic pattern of a transaction

New facts resulting from four basic events are instantiated by such classes of objects as Request, Promise, Stated Result and Accepted Result. Two interaction loops between Requester and Performer of the basic transaction pattern are composed together. A promise is created in the first interaction loop. It can be consumed in the next interaction loop. Created or terminated objects and their properties are interpreted as facts, which represent requests, promises and statements about delivered or accepted results. For instance, the Create Reservation action in figure 6 can be interpreted as a Hotel’s promise to Provide Hotel Room. Request Room and Create Reservation are typical coordination actions, which can be viewed as triggering events for a corresponding production action. Two interaction loops, which are illustrated in figure 9, represent one example of the basic pattern of a transaction. It is obvious from the presented example that the Provide Hotel Room business event is viewed as a production event. It creates effects, which can be associated with the transition P: State in the conversation schema (see figure 8). Production event creates the new fact of Stated Result. For example, an Assigned Hotel Room is supposed to create a value for a Room Guest.

148

R. Gustas and P. Gustiene

Fig. 9. Example of the basic transaction pattern

It is often the case in practice that the promise or acceptance actions are missing, because they are performed tacitly. For instance, the Create Reservation and Accept Room actions are missing in the following service interaction loop: If Request Room(Room Guest Hotel) then Provide Hotel Room(Hotel  Room Guest). The pattern, which is illustrated in figure 8, defines the case when service requester and performer are consenting to each other’s communication actions. For a communication action to be successfully performed an agent initiates the interaction flow and a recipient agrees to accept it. An enterprise system can be analyzed as the composition of the autonomous interacting components, which may not necessarily consent with each other. Actors can be involved in various interaction loops, because they want to get rid of problems or to achieve their goals. Goals, problems and opportunities [14] may help to understand why different actors act, react or not act at all. For instance, an agent may be not interested to initiate any interaction, or a recipient may refuse to accept the interaction flow. There are many other alternative business events [7], which may take be superimposed on the basic transaction pattern.

6

Alternative Interaction Loops

Alternative interaction loops should be introduced to handle possible breakdowns in the basic interaction pattern. These alternatives are represented by such reclassification actions as Reject and Decline in the standard pattern of a transaction. The alternative actions are necessary for actors involved in the business process to deal with unexpected situations. For instance, a performer may fail to deliver a desired result on time. A performer may experience difficulties in satisfying a request. For example, Hotel Reservation System may Reject Request, because the request requirements

A Method for Conceptual Modeling of Semantically Integrated Use-Case Scenarios

149

were simply incorrect or incomplete. Instead of promising, the performer may respond by rejecting request. Requester may also express disappointment in stated result and decline it. Decline is represented by the termination of Stated Result and creation of Declined Result object. For instance, the Hotel Guest may decline the assigned hotel room, which was assigned by the Provide Hotel Room action. The basic transaction pattern can be supplemented with two dissent patterns, which are represented in figure 10. This extended schema is known as the standard pattern [7].

Fig. 10. The standard transaction pattern

Alternative actions can be represented by different reclassification, creation or termination events with the same object. For example, Request can be either reclassified to Promise or to Rejection. It means that Promise and Reject actions are exclusive. The creation of Promise or creation of Rejection object can only be performed once. The alternative actions must be introduced to handle the breakdowns in the main interaction pattern. For instance, the normal Reserve Room use-case scenario can be accomplished if and only if one or more desirable types of rooms are available for the required period of stay. This flow of events would fail when there are no available rooms, which can be offered. The alternative flow is inserted when the normal flow of events fails. The Handle Waiting List use-case represents such alternative, which can be described as follows: 1) If desirable type of room is not available (failure to offer at least one available room), then the Hotel Reservation System offers waiting list possibility. 2) If customer rejects waiting, then the Hotel Reservation System declines hotel reservation request by Reject Request action. 3) If customer accepts waiting, then the system puts customer on a waiting list and preserves information about his Hotel Reservation Request.

150

R. Gustas and P. Gustiene

Two different alternatives for handling a Hotel Reservation Request are represented in figure 11 by Reject Request and Handle Waiting List actions.

Fig. 11. Two alternatives of handling Hotel Reservation Request

Create Reservation action is an event of the expected scenario. It can be performed successfully on a condition that Hotel Reservation is created. Possibility of failure to compose the Hotel Reservation of at least one Hotel Room[Reserved] would cause a breakdown in the basic transaction pattern, which requires the definition of at least one alternative event. The first alternative is represented by Handle Waiting List action, which defines the reclassification effects of Hotel Reservation Request object. The second alternative is the termination of Hotel Reservation Request by Reject Request action. This option may be caused by a failure of the Handle Waiting List action. Please note that Customer[On Waiting List] object can be created just in case the Customer agrees to Accept Waiting in the underlying interaction loop. In practice, it is also common that either requester or performer is willing to completely revoke some events. For example, the requester may withdraw his own request. There are four cancellation patterns [7], which may lead to partial or complete rollback of a transaction. Every cancellation action can be performed if the corresponding fact exists. For instance, the Withdraw Request action can be triggered, if a request was created by the Request action. In our example, Withdraw Request action is missing. Nevertheless, it is reasonable and should be introduced. The possibility to superimpose four cancellation patters on the standard pattern is not the only advantage of the presented modeling approach. The SICM method has sufficient expressive

A Method for Conceptual Modeling of Semantically Integrated Use-Case Scenarios

151

power to cover the other special cases, which are not matching the standard pattern and four cancellation patterns. For instance, it is unclear how the methodology for design and engineering of organizations [7] would cope with the Handle Waiting List alternative, which is represented in figure 11. This option is also excessive in comparison with all legal transitions, which are defined by the conversation for action schema [18].

7

Concluding Remarks

The goal of this paper was to demonstrate the advantages of conceptual modeling approach, which allows introducing evolutionary extensions of use-case scenarios. We have demonstrated how use-case narrative descriptions can be replaced by graphical representations. Integrated conceptual modeling method was used to visualize event flows in terms of underlying, sequential and alternative interaction loops, which are fundamental elements for composition of use-cases scenarios. Elementary interaction loops are important for system architects to construct scenarios, which have an understandable structure. The networks of interaction loops may span across several organizations or partnerships. Each interaction loop can be analyzed separately as it is required by the principle of separation of concerns. In such a way, interaction loops provide a natural way of decomposition of use-case scenarios. Introducing underlying interaction loops allows system designers to meet evolving needs of stakeholders and to avoid scenario breakdowns, which can be viewed as hidden requirements defects. The breakdowns in the main scenario can be eliminated by introducing the alternative actions, which are necessary to deal with failures. The presented way of interaction loop composition suggests a flexible way for managing the complexity of conceptual representations. We have demonstrated by examples some basic principles of a non-traditional conceptual modeling approach, which allows designers to visualize and to analyze semantic integrity between conceptual representations of use-case scenarios. The advantage of such conceptual representations is that interaction loops can be gradually enhanced or effectively replaced on demand. Semantic integrity problems in the early system development stages are one source of errors, because use-case descriptions touch several classes. It is very difficult to achieve semantic integrity between the static and dynamic aspects of complex scenarios, because the conventional conceptual modeling methods are developed for analyzing business processes and business data in isolation. Most graphical modeling techniques are not flexible for the visualization of interplay among behavioral, interactive aspects and structural changes of objects. It was also demonstrated how sequential, iterative, parallel and alternative behavior is captured by identifying the transition dependencies between classes in various interaction loops. Separation of concerns is achieved by decomposing complex scenarios into simple underlying loops, which can be analyzed in isolation. The presented graphical descriptions of scenarios are easier to understand, extend and maintain.

152

R. Gustas and P. Gustiene

References 1. Arlow, J., Neustadt, I.: UML 2 and the Unified Process. Practical object-oriented analysis and design. Pearson Education, Inc. (2009) 2. Cockburn, A.: Writing Effective Use Cases. Addison-Wesley, Boston (2001) 3. Miles, R., Hamilton, K.: Learning UML 2.0. O’Reilly, Sebastopol (2006) 4. Jacobson, I., Ng, P.-W.: Aspect-Oriented Software Development with Use Cases. Pearson, New Jersey (2005) 5. Denning, P.J., Medina-Mora, R.: Completing the Loops. Interfaces 25(3), 42–57 (1995) 6. Gustas, R.: Modeling Approach for Integration and Evolution of Information System Conceptualizations. International Journal of Information Systems Modeling and Design 2(1), 45–73 (2011) 7. Dietz, J.: Enterprise Ontology: Theory and Methodology, p. 256. Springer, Berlin (2006) 8. Wagner, G.: The Agent-Object-Relationship Metamodel: Towards a Unified View of State and Behavior. Information Systems 28(5) (2003) 9. Zachman, J.A.: A Framework for Information Systems Architecture. IBM System Journal 26(3), 276–292 (1987) 10. Ferrario, R., Guarino, N.: Towards an Ontological Foundation for Services Science. In: Domingue, J., Fensel, D., Traverso, P. (eds.) FIS 2008. LNCS, vol. 5468, pp. 152–169. Springer, Heidelberg (2009) 11. Gordijn, J., Akkermans, H., van Vliet, H.: Business Modelling is not Process Modelling. In: Conceptual Modeling for E-business and the Web, pp. 40–51. Springer, Berlin (2000) 12. Gustas, R.: Conceptual Modeling and Integration of Static and Dynamic Aspects of Service Architectures. In: International Workshop on Ontology, Conceptualization and Epistemology for Information Systems, Software Engineering and Service Sciences, pp. 17–32. Springer, Hammamet (2010) 13. Dennis, A., Wixom, B., Tegarden, D.: System Analysis and Design with UML. An ObjectOriented Approach, 3rd edn. John Wiley & Sons (2010) 14. Gustas, R., Gustiené, P.: Pragmatic - Driven Approach for Service-Oriented Analysis and Design. In: Johannesson, P., Söderström, E. (eds.) Information Systems Engineering: From Data Analysis to Process Networks, pp. 97–128. IGI Global, New York (2008) 15. Evermann, J., Wand, Y.: Ontology Based Object-Oriented Domain Modeling: Representing Behavior. Journal of Database Management 20(1), 48–77 (2009) 16. Bunge, M.A.: Treatise on Basic Philosophy: A World of Systems Ontology II, vol. 4. D. Reidel, Dordrecht (1979) 17. Gustas, R.: Overlaying Conceptualizations for Managing Complexity of Scenario Specifications. In: IFIP WG8.1 Working Conference on Exploring Modeling Methods for System Analysis and Design, London, UK (2011) 18. Winograd, T., Flores, F.: Understanding Computers and Cognition: A New Foundation for Design. Ablex, Norwood (1986)

Ontology-Driven Business Process Design Carla Marques Pereira2,3, Artur Caetano1,3, and Pedro Sousa1,2 1

Department of Computer Science and Engineering, Instituto Superior Técnico, Technical University of Lisbon, Av. Rovisco Pais, 1049-001 Lisboa, Portugal 2 Link, Av. Duque de Ávila, 23, 1000-138 Lisboa, Portugal 3 INESC ID & INESC INOV, Rua Alves Redol 9, 1000-029 Lisboa, Portugal [email protected], {artur.caetano,pedro.sousa}@ist.utl.pt

Abstract. Rapid business change demands the ability to adapt, rearrange and reinvent business processes while keeping the alignment with supporting information systems. However, such tasks require a business process to be consistently specified and modelled. To address this issue, this paper describes an organizational taxonomy that defines a controlled vocabulary to design business processes using the concepts of information entity, business process, organizational unit, actor, business schedule and business goal. Keywords: organizational taxonomy, ontology, business process design, conceptual modelling.

1

Introduction

Business process management plays a central role at operational, organizational and technological levels [1-3]. Business process modelling produces abstract descriptions of business processes that are a central asset to the organization as they enable its specification, documentation, analysis and engineering through multiple paradigms, languages and techniques [3-7]. However, process modelling languages are often criticised due to the lack of mechanisms to deal with domain changes and with the integration of requirements from multiple stakeholders [8, 9]. The goal of this paper is to define a shared language that enables specifying business processes while serving as a communication, analysis and discussion platform to its stakeholders. Such goal attempts to minimize having multiple inconsistent specifications of the same business process for different stakeholder subgroups. We argue that such inconsistent specifications stem from two main causes. The first one is that the different stakeholders of the same process tend to belong to different organizational areas as a business process also tends to crosscut intra- or even inter-organizational boundaries. Hence, stakeholders have contrasting concerns and thus focus on different perspectives of a process such as performance, auditing, information systems and compliance. This problem also arises during the design of architectural descriptions and can be addressed by having multiple views over the same conceptual domain [10, 11]. The second cause is that the specification of a business process is intrinsically tied to its design team. Therefore, the process design is T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 153–162, 2011. © IFIP International Federation for Information Processing 2011

154

C.M. Pereira, A. Caetano, and P. Sousa

influenced not only by the organizational factors that do specify the process but also by the team’s knowledge and background. The outcome of such observation is that two different design teams modelling the same process tend to obtain different specifications. Moreover, the formally defined modelling concepts used in widely used business processes modelling languages such as BPMN and EPC depend on natural language to specify concepts such as activities, events and information. Therefore, if the different modelling teams do not share a common vocabulary to describe a business process, the task of assessing whether two process models are actually equivalent is complex. This paper describes a controlled vocabulary that encompasses a set of core constructs to define a business process. This vocabulary is independent of the actual process modelling language and intends to facilitate the analysis and communication of a process, especially when multiple design teams or multiple stakeholders are involved. The remainder of this paper is structured as follows. The next section reviews related work. Sections 3 introduce the concepts, ontology and taxonomy to support business design. Finally, section 4 summarizes the project.

2

Related Work

Organizations require different perspectives for modelling business processes, such as modelling for compliance, documentation, process redesign or execution. Regardless of the goal, the output should be a process representation that enables its analysis, sharing and communication. Reference models and taxonomies can be used to increase the consistency of process modelling through the systematic reuse of proven and best practices across process design projects [12, 13]. Reference models, such as the Supply Chain Operations Reference (SCOR), capture knowledge about common activities, information artefacts and flows encountered in specific application domains. However, their do not go beyond pre-defined repositories of process models and provide little or no guidance to support the adaptation of these models to specific needs or different application contexts. Moreover, there is a lack of notations and methods to model reference processes and to enable their systematic reuse of reference process models in business process management projects. Techniques such as Petri nets [14], flowcharts [15], state charts [16], EPC [17], UML Activity Diagrams [18] and BPMN [19] are valuable for specifying the controlflow associated with a business process. Entity-relationship diagrams [20] and data flow diagrams [21] are useful to capture the data handled by a process. Other approaches such as Speech Act Theory [22] and the Language/Action Perspective [23] model the interaction between actors and systems and how they do so. However, most of these techniques are only able to address a single aspect of the domain, with a strong emphasis on control or data flow modelling. Therefore, supplementary modelling mechanisms are required to provide a comprehensive description of the process. This also implies that these different mechanisms need to be tightly integrated so that the resulting models are consistent and traceable. Since assessing the coherence and equivalence of the different representations of the same process is complex, organizations are compelled to design and maintain various representations of the same domain. This

Ontology-Driven Business Process Design

155

makes analysing and communicating the processes complex and also hinders the design of the information systems that support these business processes because there is no single source of knowledge. For instance, an EPC model specifies resources and objects which are required for handling specific activities at one process level but defines no criteria on how to decompose those elements. And BPMN and EPC express activities, information and events in natural language which tend lead to further modelling ambiguity if no proper shared vocabulary is defined. The next sections introduce the concepts required to define such a shared vocabulary regardless of the process modelling language.

3

Core Concepts

The purpose of business process design is the construction of concise and unambiguous models of a business or a business area in terms of the activities that take place in the organization. The fundamental assumption behind this project is that a business process can be represented through the identification of the concepts that are associated to the following six dimensions: what, where, who, when, why and how, i.e. the classic 5W1H dimensions [24, 25]. A business process can be functionally decomposed into a set of individual tasks. While we refer to a business process as a set of activities, both concepts are actually interchangeable. However, the resulting decomposition structure needs to be formalized in order to avoid unambiguity. We use Krogstie & Sølvberg’s definition of hierarchy to model the functional decomposition of a process into finer grained elements [26]. This definition states that H is an hierarchy iff it is an ordered triple H=<S,b,D> where S is a nonempty set, b a distinguished element of S and D a binary relation over S such that: 1. S has a single beginner, b. (H has one and only one supreme commander) 2. b stands in some power of D to every other member of S. (That is, no matter how low in the hierarchy an element of S may stand, it is still under the command of the beginner) 3. For any given element y of S except b, there is exactly one other element x of S such that Dxy. (i.e. every member has a single direct boss.) 4. D is transitive and anti-symmetric. Therefore, we assume the resulting process tree is strict hierarchal graph i.e. a digraph whose underlying graph is a tree, and for which there is one specific vertex from which all other vertices can be reached. The relationships between the concepts can take the form of classification, aggregation and generalization: • Classification ("is member of”). Abstracts a concept as a class of objects characterized by common properties. • Aggregation (“is part of”). Defines a new class from a set of classes that represent its component parts, i.e., thinking of concepts as wholes, not just a collection of their attributes/components.

156

C.M. Pereira, A. Caetano, and P. Sousa

• Generalization (“is a”). Defines a subset relationship between the elements of two or more classes, i.e., abstracting commonalities of several classes into a superclass.

4

The Organizational Taxonomy

An organizational taxonomy defines a controlled vocabulary which aims to be understandable to all the process stakeholders. It is a collection of terms organized as an hierarchical structure as described in the previous section. Each term is in one or more parent-child relationships to other terms in the taxonomy. The organizational taxonomy here described represents the hierarchical classification of the concepts used to represent business processes. A taxonomy helps to structure, classify, model and represent the concepts and relationships pertaining to business process design while enabling a community to come to agreement and to commit to use the same terms in the same way. We define a business process (v. Figure 1) as a set of connected activities (how) which consumes and produces tangible or intangible artefacts (what), is performed by people or systems (who), contributes to achieving goals (why), takes place in a specific location (where) and during a specific period of time (when).

Fig. 1. The six core concepts of a business process and the corresponding six classification dimensions (why, what, where, when, who and how)

For each of these concepts we create a taxonomy based on the categorization of each concept instantiation in the classification structure. The definition of the classification structure is directly dependent of the way how the organization conducts its business and it can be different from organization to organization. For this reason it is necessary to define the ontology that should be applied for each concept and this must be recognized for all the stakeholders. We consider an ontology to be a formal explicit description of the concepts within a domain of discourse, the properties of each concept and the relationships between the concepts. The following subsections describe the ontologies for each of the concepts depicted in Figure 1, namely BUSINESS PROCESS, INFORMATION ENTITY, ORGANIZATIONAL UNIT, ACTOR, BUSINESS SCHEDULE and BUSINESS GOAL.

Ontology-Driven Business Process Design

4.1

157

Business Process

A BUSINESS PROCESS is a set of connected activities that consumes and produces ttangible or intangible artefactss, is performed by people or systems, contributes to achiieving goals, takes place in a specific location and during a specific period of timee. A business process can be fun nctionally decomposed as a set of individual tasks. Thuus, a business process: • • • • •

MATION ENTITIES as input/output. Has one or more INFORM Takes place in one or more m ORGANIZATION UNITS. Is performed by one or more m ACTORS. Occurs in a specific BUSSINESS SCHEDULE. Contributes to one or more BUSINESS GOALS.

Abusiness process can be cllassified as: • MACRO-PROCESS. A laarge-scale business PROCESS that is initiated by a custom mer request, or by the decisiion of the company to enter a new line of business, andd results in the delivery of a process or service to a customer, i.e., purchase, manuffacturing, logistics, custom mer service, marketing, sales, etc. • PROCESS. At its most generic, g any set of ACTIVITIES performed by a business tthat is initiated by an event, transforms information, materials or business comm mitments, and produces an output. Process can be decomposed indefinitely. • ACTIVITY. The leaves of the process tree are activities. So, an ACTIVITY iis a PROCESS that cannot be further decomposed. The following diagram dep picts an ontology for a business process according to the concepts of business processs, macro-process, process and activity.

Fig. 2. App plication of the ontology to a business process

158

4.2

C.M. Pereira, A. Caeetano, and P. Sousa

Information Entity y

An entity is any person, plaace, concept, thing, or event that has meaning in the conttext of the business, and about which w data may be stored [27]. INFORMATION ENTITIES are composed by several attrib butes and can have relationships with other entities. An entity can be classified as: • THING. It is an information “bag” that represents the many different types of ellectronic or physical artefaacts that are important to the business, such as documeents, products, resources, etc.. • DATA ENTITY. Is sometthing that has some meaning in the context of the business, and about which data may m be stored. The instances of data type represent a sspecialization of a THING, such s as invoice, order, application form, receipt, custom mer, supplier, employee, etc. • PROPERTY. A relevant characteristic of a DATA ENTITY that further describes w what that data entity is in thee context of the business, such as order number, item prrice, customer address, invoicce date, etc.

Fig. 3. Appllication of the ontology to an information entity

4.3

Organizational Uniit

The organizational structurre includes information about the ORGANIZATIONAL UN NITS that make up an organizatio on, the human resources that belong to those organizatioonal units, as well as the structture and relationships that connect them all together. T The three following concepts claassify organizational unit:

Ontology-Driven Business Process Design

159

• GEOGRAPHICAL LOCATION. Place where the organization is located which may be a country, city, geographical area, etc. • PHYSICAL LOCATION. Building where the organization is installed, e.g. headquarters, delegation, office, store, etc. • UNIT. Used to create a hierarchy of organizational areas within a organization, e.g. division, department, section, etc. 4.4

Actor

Actor specifies a role played by a user or any other system that interacts with the subject. ACTORS may represent roles played by human users, information systems, or other active subjects. Note that an actor does not necessarily represent a specific physical entity but merely a particular role of some entity that is relevant to the specification of its associated use cases. Thus, a single physical instance may play the role of several different actors and, conversely, a given actor may be played by multiple different instances [28, 29]. The concept of actor is classified as: • POSITION. A function within a chain of command of an organization that has the responsibility for decisions involving the use of the organizational resources, i.e., manager, chief business officer, clerk, supervisor, project manager, etc. 4.5

Business Schedule

A BUSINESS SCHEDULE is a plan that specifies time periods for completing specific activities. In the context of this work, business schedule is the set of events that are important for the enterprise and have business process associated to them. A schedule is classified through events: • EVENT. All occurrences happening at a determinable time that the organization must be aware of. Event can be decomposed as other events. 4.6

Business Goal

BUSINESS GOALS are a set of objectives that are accomplished by one more business process. Goals are classified as: • GOAL. An end which the organization seeks to achieve through its operations. • OBJECTIVE. A decomposition of GOAL that can be achieved within a defined timeframe and a set of resources. • KPI. Key performance indicators can be associated to GOALS and OBJECTIVES to provide measures that will reflect success factors, such as time to answer a customer request or maximum lead time of a process. 4.7

Relationships between the Taxonomy Concepts

Table 1 summarizes the relationships between business processes and the remaining concepts.

160

C.M. Pereira, A. Caetano, and P. Sousa Table 1. Relationships between the core concepts and a business process

Concept

Macro-process Thing

Information Entity

Process

Activity

X

X

X

Data Entity

X

Property Organizational Unit

Geographical location

X

Physical location

X

Unit

X X

X

Actor

Position

X

X

X

Business Schedule

Event

X

X

X

Goal

X X

X

Business Goal

Objective KPI

5

X

X

Conclusions

The lack of a common language between the stakeholders and the process designers results in a significant gap between different modelling perspectives. To reduce this gap, this paper has proposed a controlled vocabulary to support business process design. This vocabulary is grounded on six dimensions of inquiry (how, where, why, when, what, who). The concepts pertaining to the vocabulary (information entity, business process, organizational unit, actor, business schedule and business goal) are organized as a taxonomy that allows for the hierarchical creation of an ontology that describes the specific domain of the organization. This approach has been experimented and validated in several professional projects at Link (www.link.pt). Acknowledgments. The work described in this paper was co-supported with funding from the National Strategic Reference Framework – Quadro de Referência Estratégico Nacional, project QREN 6652.

References 1. Davenport, T., Short, J.: The New Industrial Engineering: Information Technology and Business Process Redesign. Sloan Management Review 32(5), 554–571 (1990) 2. Hammer, M., Champy, J.: Reengineering the Corporation: A Manifesto for Business Revolution. Nicholas Brealey Publishing, London (2001) 3. Dietz, J.: Enterprise Ontology: Theory and Methodology, p. 244. Springer, New York (2006) 4. van der Aalst, W.M.P., ter Hofstede, A.H.M., Weske, M.: Business Process Management: A Survey. In: van der Aalst, W.M.P., ter Hofstede, A.H.M., Weske, M. (eds.) BPM 2003. LNCS, vol. 2678, pp. 1–12. Springer, Heidelberg (2003) 5. OMG Business Process Model and Notation (BPMN), version 2.0 (2011)

Ontology-Driven Business Process Design

161

6. Ko, R., Lee, S., Lee, E.: Business process management standards: a survey. Business Process Management Journal 15(5) (2009) 7. Russell, N., van der Aalst, W.M.P., ter Hofstede, A.H.M., Edmond, D.: Workflow Resource Patterns: Identification, Representation and Tool Support. In: Pastor, Ó., Falcão e Cunha, J. (eds.) CAiSE 2005. LNCS, vol. 3520, pp. 216–232. Springer, Heidelberg (2005) 8. Dumas, M., Hofstede, A.H.t., Aalst, W.v.d.: Process Aware Information Systems: Bridging People and Software Through Process Technology. Wiley Publishing (2005) 9. Ellis, C.A., Nutt, G.J.: Workflow: The Process Spectrum. In: NSF Workshop on Workflow and Process Automation in Information Systems: State-of-the-Art and Future Directions, Athens, GA (1996) 10. IEEE Computer Society, IEEE Std 1471-2000: IEEE Recommended Practice for Architecture Description of Software-Intensive Systems. IEEE, New York (2000) 11. Davis, P., Tolk, A.: Observations on New Developments in Composability and MultiResolution Modeling. In: Winter Simulation Conference WSC 2007, Washington DC, USA (2007) 12. Malone, T.W., Crowston, K., Herman, G.A.: Organizing Business Knowledge: The MIT Process Handbook. MIT Press, Cambridge (2003) 13. Council, S.-C.: Supply Chain Operations Reference, SCOR (2003), http://supplychain.org/ 14. Petri, C.A.: Kommunikation mit Automaten. In: Institut für instrumentelle Mathematik. University of Bonn, Bonn (1962) 15. Schriber, T.J.: Fundamentals of Flowcharting. Wiley, New York (1969) 16. Harel, D.: Statecharts: A visual formalism for complex systems. Sci. Comput. Program. 8(3), 231–274 (1987) 17. Keller, G., Nüttgens, M., Scheer, A.-W.: Ereignisgesteuerter Prozessketten (EPK). In: Semantische Prozessmodellierung auf der Grundlage, Saarbrücken, Germany (1992) 18. OMG, Unified Modeling Language Specification: Superstructure, version 2.0, Revised Final Adopted Specification (ptc/04-10-02). Object Management Group (2004) 19. OMG, Business Process Modeling Notation (BPMN) Specification. v 1.1 (formal/200801-17) (January 2008) 20. Chen, P.: The entity-relationship model: Towards a unified view of data. ACM Transactions on Database Systems 1(1) (1976) 21. Gane, C., Sarson, T.: Structured Systems Analysis: Tools and techniques. Prentice-Hall, Englewood Cliffs (1979) 22. Searle, J.: Speech Acts: An Essay in the Philosophy of Language. Cambridge University Press, Cambridge (1969) 23. Winograd, T.: A language/action perspective on the design of cooperative work. HumanComputer Interaction 3(1), 3–30 (1988) 24. Sousa, P., et al.: Applying the Zachman Framework Dimensions to Support Business Process Modeling. In: Digital Enterprise Technology Perspectives and Future Challenges 2007, pp. 359–366. Springer, US (2007) 25. Sousa, P., Lima, J., Sampaio, A., Pereira, C.: An Approach for Creating and Managing Enterprise Blueprints: A Case for IT Blueprints. In: Albani, A., Barjis, J., Dietz, J.L.G. (eds.) CIAO! 2009. LNBIP, vol. 34, pp. 70–84. Springer, Heidelberg (2009) 26. Krogstie, J., Sølvberg, A.: Information Systems Engineering - Conceptual Modeling in a Quality Perspective. Kompendiumforlaget, Trondheim (2003) 27. Spewak, S., Steven, H.: Enterprise Architecture Planning: Developing a Blueprint for Data, Applications and Technology. Wiley-QED Publication, New Jersey (1992)

162

C.M. Pereira, A. Caetano, and P. Sousa

28. List, B., Korherr, B.: A UML 2 Profile for Business Process Modelling. In: Akoka, J., Liddle, S.W., Song, I.-Y., Bertolotto, M., Comyn-Wattiau, I., van den Heuvel, W.-J., Kolp, M., Trujillo, J., Kop, C., Mayr, H.C. (eds.) ER Workshops 2005. LNCS, vol. 3770, pp. 85– 96. Springer, Heidelberg (2005) 29. Caetano, A., Silva, A.R., Tribolet, J.: Business Process Model Decomposition with Separation of Concerns. In: 25th Annual ACM Symposium on Applied Computing, ACM SAC 2010. ACM, Sierre (2010)

On the Use of a Role Ontology to Consistently Design Business Processes Artur Caetano1,2 and José Tribolet1,2 1

Department of Computer Science and Engineering, Instituto Superior Técnico, Technical University of Lisbon, Av. Rovisco Pais, 1049-001 Lisboa, Portugal 2 INESC ID & INESC INOV, Rua Alves Redol 9, 1000-029 Lisboa, Portugal {artur.caetano,jose.tribolet}@ist.utl.pt

Abstract. The functional decomposition of a business process breaks it down into progressively less granular activities. Decomposition contributes to the modular design of a system, the reuse of its parts and to its overall comprehensibility. But achieving these qualities requires a business process to be decomposed consistently, which implies it is always split into an identical set of activities according to a specific purpose, regardless of the modeller’s and modelling context. This paper describes an application of the principle of role-based separation of concerns to consistently decompose a business process into its constituent atomic activities, thus separating its distinct features and minimizing behaviour overlap. An activity is abstracted as collaboration between role types that are played by entities. The decomposition method successively separates the overlapping roles until an activity is specified as a collaboration of an orthogonal set of role types. The method facilitates the consistent decomposition of a business process and the identification of its atomic activities. The relevance of the method is assessed through a number of scenarios according to the guidelines of design science research. Keywords: business process modelling, functional decomposition, separation of concerns, enterprise architecture.

1

Introduction

It is widely accepted that one of the fundamental problems in the design and development of knowledge-based systems is extracting information from the experts and then translating it to the form of some knowledge base in order to attain a given purpose. As in the case of business process modelling, this transformation is not straightforward as the source knowledge is often not structured or formalized and tends to be of complex nature. Furthermore, the purpose of the model itself may not be well defined or understood by all of its stakeholders. As a matter of fact, a number of researchers posit that complexity is an essential property of design activities in general due, in part, to the inevitably incomplete formulation of the problem and to our inability to cope simultaneously with all of the constraints of a given problem. Service-oriented architecture is an architectural style for constructing systems from a set of universally interconnected and interdependent services. A service is a unit of T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 163–176, 2011. © IFIP International Federation for Information Processing 2011

164

A. Caetano and J. Tribolet

functionality that some entity makes available to its environment. This style of architecture promotes reuse at a macroscopic service level and can simplify the usage and interconnection the business, application and technological assets within and across organizations [1]. Service orientation promotes a layered view of an enterprise architecture’s models. Service layers provide functionality to higher layers and are realized in lower implementation layers. For instance, the ArchiMate enterprise modelling language [2] defines three layers: the business layer defines which products are offered to external customers through business processes; the intermediate application layer supports the business layer with application services which are realised by application components; finally, the technology layer offers infrastructural services needed to run applications. A business process is a set of interrelated value-adding activities [3]. Activities are often modelled as opaque transformation functions that map inputs to outputs. This abstraction strategy models an activity as a black-box and focus on its external behaviour. The resulting models conceptually divide a business system into a hierarchy of functions [4]. Thus, functionally decomposing a business process entails its recursive separation into a set of more detailed activities. Business process models translate the knowledge about how an organization operates. These models are fundamental to enterprise architecture as they support the communication, analysis, implementation and execution of the organization’s structure, business, systems and technology [2, 5]. Process models also provide the means to analyze alternative designs intended to align the business with the information systems and technology. However, the process modelling must cope with the multiple views and goals of the different organizational stakeholders. Moreover, the elicitation, modelling and analysis of the processes of an organization is often the result of merging the partial contributions of different teams, probably with different backgrounds and experience. Put together, these factors lead to models that lack consistency. Examples of inconsistency include using different levels of modelling detail and the incoherent naming of the activities and entities of a process. Inconsistent process models are not only hard for their users to understand but also hamper the tasks of process analysis and redesign as they may leave out relevant information or lead to erroneous or ambiguous interpretations of the process content. Such inconsistency also negatively contributes to the identification and the specification of the services that are required to support the process. Consistent business process decomposition can significantly improve the clarity and the overall model integrity as well as minimizing the omission of relevant information [6]. Decomposition is also a means to modularize large systems and to facilitate the reuse of partial models and favours the compactness of a specification as it allows multiple levels of detail to co-exist and coupling to be reduced [7]. As a consequence of abstraction, models become easier to understand and communicate, which, in turn, make their validation, redesign and optimization easier. This paper proposes using the separation of concerns principle to facilitate the consistent decomposition of a business process and the unambiguous identification of its atomic activities thus contributing to the task of identifying the supporting services. To do so, we present a method that specifies how to decompose a business process according to the concerns that are involved in the specification of its activities.

On the Use of a Role Ontology to Consistently Design Business Processes

165

The remainder of this paper is structured as follows. The next section reviews related work. Section 3 introduces the concepts of natural type, role type and activity. Sections 4 and 5 describe the functional decomposition method and the underlying role ontology along with a running example. Finally, section 6 summarizes the research methodology and section 7 summarizes the proposal and provides an outlook on future work.

2

Related Work

Functional decomposition is supported at language level by most process modelling languages, including ArchiMate [2], BPMN [8], EPC [9] and IDEF-0 and IDEF-3 [10]. The decomposition of subsystems through the hierarchic classification of process models has also been applied to Petri nets [11] and BPEL [12]. Although these approaches make possible creating a hierarchical representation of a process, their intent is not the definition of techniques for consistent activity decomposition but, instead, the representation of generic decomposition structures. Nevertheless, the shortcomings of the lack of consistency in process decomposition and in the identification of its atomic activities are pointed out by several authors [13, 14]. Several top-down decomposition approaches exploit reference models to describe how a process is structured as a hierarchy of activities. For instance, the Supply-Chain Operations Reference model describes three levels of detail to assist the definition and configuration of an organization’s supply chain [15]. The Process Clarification Framework defines a hierarchical (and static) decomposition of business processes which is 3 to 4 levels deep and crosses 12 operating and management categories [16]. Other approaches, such as the ARIS framework [9], describe processes as chains of events and tasks and prescribe the levels of detail for decomposition. The first two decomposition levels address the business viewpoint of the model, the next 3 to 4 levels focus on the structure of process operation and the lower level describes the procedural details of the tasks. However, the contents of these levels of detail are actually arbitrary. An alternative avenue of research relies on algorithmic methods to analyse the process specification and assess its consistency. One of these methods uses similarity measures derived from the syntactic and structural features of the process to detect inconsistencies between its activities [17]. These measures make use of a linguistic ontology to evaluate the similarity between the names of the activities thus assisting the detection of decomposition anomalies. Process mining techniques extract information from existing event logs and enable the discovery of business processes [18]. These bottom-up mining techniques support the verification of the conformance of a model derived from an event log against an existing model as well as identifying the atomic activities of a process [19]. Other approaches that use ontologies to specify business processes (e.g. [20-22]) also lack the means to identify atomic activities and to consistently decompose a process. Altogether, and to the best of our knowledge, existing approaches do not define the necessary means to consistently decompose a business process and to unambiguously identify the atomic activities that constitute it. The primary goal of this paper is therefore to provide a contribution to this research subject.

166

3

A. Caetano and J. Tribolet

Fundamental Concepts

Role modelling is a separation of concerns technique that is used in multiple areas of knowledge such as data modelling [23], object-oriented and conceptual modelling [24-26], framework design [27], business process modelling [28, 29] and enterprise architecture [20, 30-34]. With role-based business process, modelling an activity (a business verb) is abstracted as a set of collaborations between entities (business nouns). The entities represent the things that are of interest within a specific modelling context. Each unit of behaviour of an entity is abstracted as a role and, as a result, activities are defined by a role collaboration pattern. If no roles are being played within the system then there are no collaborations and, therefore, no activities to be modelled. Figure 1 shows the relationships and cardinalities between four entities involved in the assemble product process which we will use as a running example to illustrate the concepts outlined above. The activity assemble product is defined by the collaboration pattern between the roles being played by the entities part, assembling machine, product and person. The activity describes how a product is assembled from a number of parts by means of an assembling machine. The activity is semi-automated as the machine is operated by a person.

Fig. 1. Relationships between entities

Fig. 2. Role-based specification of the assemble product activity

Figure 2 shows the relationships between the entities result in one collaboration context where a natural type displays a specific behaviour [33, 34]. Such behaviour is abstracted as a role type. Thus, in the first collaboration context each part plays the role of input resource in their relationship with the assembling machine which, in its turn, is playing the actor role. In other context the assembling machine produces the assembled product, i.e. the product is the output resource of this actor. Finally, the person relates to the machine as its actor. The collaboration between these four roles uniquely defines the assemble product activity as depicted in Figure 2. The actor

On the Use of a Role Ontology to Consistently Design Business Processes

167

role states that an entity is able to perform some action in the context of an activity. The resource role states that an entity which is playing it can be used or consumed (input resource) or created (output resource) during the performance of an activity. The remainder of this section details the concept of entity (or natural type), role (or role type) and activity. 3.1

Natural Types and Role Types

Sowa [35] distinguished between natural types “that relate to the essence of the entities” and role types “that depend on an accidental relationship to some other entity”. By developing Sowa's ideas further, Guarino presented an ontological distinction between these two types [36]. This distinction is based on the concepts of foundedness and semantic rigidness. A type is considered founded if its specification implies a dependency or relation to some other individual. A type is semantically rigid if the identity of an individual depends on the type assigned to it. If the type is removed from the individual then it cannot be further identified nor classified. Thus, a type is not semantically rigid if it can be assigned to and removed from an individual without changing its identity. Based on the above, a type that is both founded and not semantically rigid is a role type. In contrast, a natural type is characterized by being semantically rigid and not founded. To illustrate the above classification properties, let us take the example of Figure 2 and classify the concepts of person and actor as either natural or role types. First, let us focus on the “foundedness” of these concepts. Actor is a founded type since for something or someone to be assigned the actor type there must be something being acted upon. Conversely, the person type is not founded since it exists on its own terms. It defines the identity of the individual to which it is assigned to, regardless of its relationships with any other individual. Thus, the person type is not founded whereas the actor type is founded. Regarding “semantic rigidness”, the actor type is not semantically rigid because its identity is independent of the individual to whom the type is assigned to. This means the actor type is not able to identify the individual by itself. On the other hand, the person type is semantically rigid as its identity is directly coupled to the individual’s identity. Therefore, actor is a role type (founded and not semantically rigid) whereas person is a natural type (not founded and semantically rigid). Natural Types Entities are natural types. In enterprise modelling, an entity describes a thing that an organization deems relevant to specify in order to fulfil the purpose of a model. Entities model concepts such as persons, places, machines, resources, contracts and products. According to the definition of natural type, an entity can be unambiguously identified and defined in isolation, i.e. without any relationship with other types. Entities can be classified according to its intrinsic features. Entities may relate structurally to other entities (e.g. an order is composed of items).

168

A. Caetano and J. Tribolet

Role Types A role type, or role for short, is the observable behaviour of an entity in the scope of a specific collaboration. Different roles separate the different concerns that arise from the collaborations between entities. Hence, a role represents the external visible features of that entity when it collaborates with another entity in the context of an activity. An entity relates to other roles through the play relationship. An entity that plays no roles is not participating in any activity since it is not able to produce actual behaviour. An entity enters the role when it starts playing it and leaves the role when the specific behaviour specified by the role is concluded. Each role adds a set of external features to an entity in the context of that collaboration. This effectively separates the entity’s feature space since its intrinsic features are differentiated from each of the external features that transiently relate to an entity through the roles it plays. Activities A business process is an ordered execution of activities that produces goods or provides services that add value to the organization's environment or to the organization itself. Thus, modelling a business process involves specifying the set of activities that define its operation and the flow that defines how the activities are coordinated. An activity is specified by a collaboration of role types. It is a behaviour element that describes part of the functionality available to the organization. Since a role type separates the description of the intrinsic features of an entity from the features that derive from the collaborations it participates in, the specification of an activity itself is independent of the specification of the entities playing the roles. Figure 2 depicts the assemble product activity as a unit of functionality that result from the collaboration between a set of roles. However, this activity model is conceptual as it may have been specified from a different perspective or with a different level of detail, which would have implied using a different role ontology. The granularity level of the activities is also arbitrary as it is always possible to add more detail to its specification. Hence, the naming of an activity is actually irrelevant for the purpose of its specification as the role collaboration pattern is the only means to specify it unambiguously. Therefore, an activity is uniquely identified by the collaboration of roles that are involved in its specification. Two activities are deemed equivalent if and only if they share the same set of role collaborations.

4

Functional Decomposition

The functional decomposition of a business process yields a set of sub-activities, each of which can be further decomposed. The behaviour of a whole process can then be constructed upwards from the lowest level of decomposition towards the top-level activity. The lowest level of decomposition describes primitive or atomic activities that cannot be further divided. The related literature (cf. section 2) describes different approaches to the functional decomposition of processes but, to the best of our knowledge, existing

On the Use of a Role Ontology to Consistently Design Business Processes

169

approaches do not provide the means to unambiguously identify what makes an atomic activity nor the mechanisms that provide consistent decomposition results. The approach proposed in this paper is to use role types as the criteria for process decomposition. This means each decomposition step separates a different concern (i.e. a role type) from the other concerns that specify the activity. An activity is deemed atomic, meaning it cannot be further decomposed, when all of its concerns are effectively separated. This translates to having no overlapping role types in the activity’s specification. It also implies that the classification of an activity as atomic actually depends on the role ontology that is being utilized to generate the process model. So, different role ontologies yield different decomposition criteria and, thus, different process models. decompose(S, R) decompose’ (S, R, D, level) D ← ∅ if R • ∅ then decompose’ (S, R, D, 1) R0 ← firstElementOf(R) decompose ← D Dlevel ← ∅ end if numInstancesOfType(RO, S) > 1 then for all r ∈ R0 do Sd ← (S – R0) ∪ r Dlevel ← Dlevel ∪ Sd decompose’ (Sd, R–R0, D, level+1) end for else decompose’ (S, R – R0, D, level+1) end if D ← D ∪ { Dlevel } end if end

The algorithm decompose (S, R) recursively separates an activity into sub-activities as long as there are overlapping concerns. S is the ordered set of all the roles type instances used in activity to be decomposed. The set R (which is a subset of the types of S) contains the role types that define the domain to be used to decompose the activity. If all the role types in S are included in R then all roles will be separated. The role types not included in R will remain overlapped after the decomposition. The output of decompose (S, R) is a set of sets. Each of these sets represents an activity, with the outer set representing the first level of decomposition. The symbol level identifies the current decomposition level with 0 representing the top level activity. The symbol D represents the output set of the decomposition and Dlevel is the set of decomposed activities pertaining to a given level of depth. The algorithm makes use of two additional functions not detailed here: firstElementOf(X) returns the first element of the set X; countInstancesOfType(t, X) counts the number of instances of the type t within the set X.

170

A. Caetano and J. Tribolet

Fig. 3. Activity A1 according to roles R1, R2, R3

Figure 3 illustrates an application of the decompose function to activity A1. A1 is defined by the collaboration of role types R1, R2, R3. Let us consider that A1 is specified by S = {a:R1, b:R1, c:R2, d:R3, e:R3} and that S maps to three role types, R = {R1, R2, R3}. Using decompose(S, R) to decompose A1 according to (R1, R2, R3), results in D = {D1, D2}. D1 is the first level of decomposition and divides A1 into {(a:R1, c:R2, d:R3, e:R3), (b:R1, c:R2, d:R3, e:R3)}. D2 is the lowest level of decomposition and comprises four atomic activities: {(a:R1, c:R2, d:R3), (a:R1, c:R2, e:R3), (b:R1, c:R2, d:R3), (b:R1, c:R2, e:R3)}. If we define the role ontology R1, R2, R3 to describe locations, goals and actors, so that R1 stands for the Locator role, which describes a geographical location, R2 is the Goal role, that models the intended state of the affairs to be achieved after executing the activity, and that R3 is the Actor role, which describes the action of someone of something operating in the context of the activity A1, we would get the model depicted on Figure 4. Decomposing A1 according to the Locator role (R1) yields two activities, A1.1 and A1.2, as shown in Figure 5. Each of these functionally separate A1 according to geographical location concern. Decomposing A1 according to the Actor role (R3) produces two activities, each focusing on the specific operations of the actor involved in A1. Note that A1 cannot be decomposed according to the Goal role (R2) as this concern does not overlap with any other role of the same type. Activities A1.1 and A1.2 can be further separated as shown in Figure 7 and 7. The decomposition of A1 according to the role tuple (Locator, Actor, Goal) results in four atomic activities, each focusing on a different concern: A1.1.1 (Office:Locator, Person:Actor, Goal:Goal), A1.1.2 (Factory:Locator, Person:Actor, Goal:Goal), A1.2.1 (Office:Locator, Machine:Actor, Goal:Goal), A1.2.2 (Factory:Locator, Machine:Actor, Goal:Goal). Note that A1 cannot be fur-

ther decomposed according to these three roles. Further decomposition is only possible if new roles are added to the ontology or additional overlapping concerns are included in the specification of A1.

Fig. 4. Activity A1 and Actor, Locator and Goal role types

On the Use of a Role Ontology to Consistently Design Business Processes

171

Fig. 5. Decomposition of activity A1 on role R1 (Locator)

Fig. 6. Decomposition of A1.1 on role R

Fig. 7. Decomposition of A1.2 on role R

This approach is unambiguous as each level of decomposition can be systematically reproduced. A business process can always be consistently separated into its constituent atomic activities and the corresponding supporting services identified. Additionally, the condition for activity decomposition is explicit as the procedure stops whenever the concerns of an activity are effectively separated. Thus, consistent process decomposition promotes service identification and reuse.

5

Role Ontology

The decomposition method relies on the specification of a role type ontology. An ontology is a formal representation of a set of concepts within a domain and the relationships between those concepts. In this particular case, the ontology represents the set of role types required to model a specific domain and the possible collaborations between these role types. A business process can be modelled from different perspectives according to the model’s goals and purpose as defined by its stakeholders. Although there are multiple classification schemes to categorize the modelling perspectives, these often crosscut the six orthogonal linguistic interrogatives (how, what, where, who, when, why). These interrogatives can be used to construct four basic modelling perspectives [37, 38]. The functional perspective represents what activities are being performed in the

172

A. Caetano and J. Tribolet

context of a given process. The informational perspective represents what informational entities (i.e. data or resources) are being manipulated by the activities of a process. The behavioural perspective represents when activities are performed and how they are performed, usually through the specification of the process orchestration. Finally, the organizational perspective represents why an activity is being performed, where it is performed and by whom. The remainder of this section exemplifies a set of roles types that addresses the above concerns according to the six interrogatives. We emphasize that the role ontology should be specified according to the requirements of the stakeholders and to the specific domain being modelled. Actor (Who) The actor role represents the action of an entity that does some task in the context of an activity. Actors are played by entities which represent people, computer systems, mechanical tools or any other devices that produce active change within an organization. A specialization scheme of the actor role type focuses on its nature, such as: social actor (people or organizations), application actor (computational or noncomputational applications that are used to perform a task) and infrastructure actor (computer hardware, machines and other devices that support the application and social actors). Another specialization scheme, which is orthogonal to the actor’s nature, includes roles such as operator, auditor and supervisor. Using the actor role as the criterion for decomposition identifies atomic that describe the actions of each individual actor. The decomposition of the assemble product activity in Figure 2 according to the actor role identifies two activities: one for the actions being performed by the person and other for the actions of the machine. Resource (What) A resource is the role played by an entity when manipulated by an actor in the context of an activity. A resource specialization scheme that focus on how a resource is transformed within an activity consists of two roles: input resource role and output resource role. The former can be further specialized as consumed resource role and used resource role, whereas the latter can be specialized as created resource role and refined resource role. Other orthogonal schemes are possible, such as classifying a resource according to its existence (e.g. tangible, intangible, etc.) Locator (Where) The locator role captures the geographical or the logical location of an entity. The sub-activities of an activity that is decomposed according to the locator role are operated in different locations. Goal, Rule (Why) A goal represents a measurable state of affairs that the organization intends to achieve. The entity plays the goal specifier role which relates to the goal fulfiller role. Goals are usually achieved by the entities playing the actor or resource role.

On the Use of a Role Ontology to Consistently Design Business Processes

173

A rule asserts conditions or operating parameters that an activity must comply with. The entity that specifies the constraint plays the rule specifier role which relates to the rule complier role. Starter, Finisher (How, When) The behavioural perspective can be captured through the starter and finisher roles. The first models the event that triggers the start of an activity while the second signals its completion. These two roles can be used to describe how the activities of a process are orchestrated, as described in the next section.

6

Research Methodology

The methodology behind the results reported in this paper is grounded on design science [39, 40]. Design science focuses on the development of solutions for practical problems. This contrasts with the development and the verification of theories as in behavioural science methodologies. Research on enterprise architecture, modelling and engineering fits the design science paradigm as its focal goal is not building information systems but creating methods and techniques to analyze, model, and understand the horizontal and vertical interfaces between the business, systems and technology [41]. The essential tangible result of a design science project consists in creating an artefact that addresses a particular issue that is relevant to a certain group of stakeholders. In this context, Hevner et al. proposed a set of guidelines to conducting design science projects [40]. The following points briefly summarize how these were applied to this work: • Design as an artefact. This project deals with applying the principle of separation of concerns to business process modelling. This paper describes an artefact that deals with business process decomposition role modelling as a separation of concerns mechanism. • Problem relevance. The artefact enables the consistent decomposition of a business process. By doing so, it addresses several problems that are relevant in enterprise engineering in general and business process modelling in particular. We emphasize the following problems: (1) how to systematically identify the atomic activities of a process; (2) how to make explicit the principles behind process decomposition; (3) how to make decomposition dependent on the specification of the process and not on the modelling team experience • Design evaluation. This paper makes use of a scenario [40] built around the artefact to demonstrate its application and utility. • Research contributions. The paper describes an algorithm for consistent business process decomposition and its applicability to the identification of business services. • Research rigour. The artefact addresses a problem identified in the enterprise engineering and business process modelling literature. The solution is grounded on the principles of role modelling, separation of concerns and business process modelling.

174

A. Caetano and J. Tribolet

• Communication of research. The research is reported through publications aimed at the practitioners and researchers within the enterprise engineering area and mainly targets business process modellers.

7

Conclusion and Future Work

Activity decomposition is an abstraction technique that enables the modularization of business processes. A decomposed process is easier to understand as each decomposition step incrementally reduces the number of overlapping concerns. This fosters the reuse and identification of the supporting services and increases the ability to communicate and analyse them. Each decomposition step provides a consistent level of detail so that the set of atomic activities comprising the lowest level of decomposition are always coherent, regardless of the stakeholder’s requirements and the modelling team’s experience. The aim of the project is to guide the procedure of process decomposition so that decompositions are explicit and consistent. The proposed method supports the decomposition of business processes according to the separation of overlapping concerns. Business processes are modelled as the collaboration of natural types that play role types in the context of activities. A role ontology is used to specify the domain of role types and constrains the decomposition space. This approach facilitates the consistent decomposition of a process and the identification of the atomic activities, which contributes to service identification. However, the scenario presented in this paper does not evaluate the impact of the specification of the ontology and the overhead introduced by role-modelling in business process modelling. To overcome this limitation, we are currently developing a set of case studies intended to evaluate the applicability of the method to large-scale business processes.

References 1. Bieberstein, N., Bose, S., Fiammante, M., Jones, K., Shah, R.: Service-Oriented Architecture (SOA) Compass: Business Value, Planning, and Enterprise Roadmap. IBM Press, New York (2005) 2. Lankhorst, M.: Enterprise Architecture at Work: Modelling, Communication and Analysis. Springer, Heidelberg (2006) 3. Davenport, T.: Process Innovation: Reengineering Work Through Information Technology. Harvard Business School Press, Boston (1993) 4. Dietz, J.L.G.: Enterprise Ontology: Theory and Methodology. Springer, Berlin (2006) 5. Op’t Land, M., Proper, E., Waage, M., Cloo, J., Steghuis, C.: Enterprise Architecture: Creating Value by Informed Governance. Springer, Heidelberg (2009) 6. Huber, P., Jensen, K., Shapiro, R.M.: Hierarchies in Coloured Petri Nets. In: Rozenberg, G. (ed.) APN 1990. LNCS, vol. 483, pp. 313–341. Springer, Heidelberg (1991) 7. Bass, L., Clements, P., Kazman, R.: Software Architecture in Practice. Addison-Wesley, Reading (1998) 8. OMG, Business Process Modeling Notation Specification. v 1.1 (formal/2008-01-17) (2008)

On the Use of a Role Ontology to Consistently Design Business Processes

175

9. Scheer, A.-W.: Business Process Modeling, 3rd edn. Springer, Berlin (2000) 10. Mayer, R.J., Menzel, C.P., Painter, M.K., deWitte, P.S., Blinn, T., Perakath, B.: Information Integration for Concurrent Engineering - IDEF3: Knowledge Based Systems Inc. (1995) 11. Reisig, W., Rozenberg, G.: APN 1998, 1st edn. LNCS, vol. 1491. Springer, Heidelberg (1998) 12. Kloppmann, M., Koenig, D., Leymann, F., Pfau, G., Rickayzen, A., van Riegen, C., Schmidt, P., Trickovic, I.: WS-BPEL Extension for Subprocesses BPEL-SPE. IBM and SAP Joint White Paper (2005) 13. Davis, R., Brabdänder, E.: ARIS Design Platform. Springer, London (2007) 14. Ingvaldsen, J.E., Gulla, J.A.: Model Based Business Process Mining. Journal of Information Systems Management 23 (2006) 15. Bolstorff, P., Rosenbaum, R.: Supply Chain Excellence: A Handbook for Dramatic Improvement Using the SCOR Model, 2nd edn. Springer, Berlin (2008) 16. APQC, APQC Process Clarification Framework - Consumer Products, version 5.0.2, 10/04/2008 (2008) 17. Hornung, T., Koschmider, A., Lausen, G.: Recommendation Based Process Modeling Support: Method and User Experience. In: Li, Q., Spaccapietra, S., Yu, E., Olivé, A. (eds.) ER 2008. LNCS, vol. 5231, pp. 265–278. Springer, Heidelberg (2008) 18. van der Aalst, W.M.P., de Beer, H.T., van Dongen, B.F.: Process Mining and Verification of Properties: An Approach Based on Temporal Logic. In: Meersman, R. (ed.) OTM 2005. LNCS, vol. 3760, pp. 130–147. Springer, Heidelberg (2005) 19. van der Aalst, W., Reijers, H., Weijters, A., van Dongen, B., de Medeiros, A.A., Song, M., Verbeek, H.: Business Process Mining: An Industrial Application. Information Systems Journal 32, 713–732 (2007) 20. Uschold, M., King, M., Moralee, S., Zorgios, Y.: The Enterprise Ontology. The Knowledge Engineering Review 13, 31–89 (2000) 21. Greco, G., Guzzo, A., Pontieri, L., Sacca, D.: An ontology-driven process modeling framework. In: Galindo, F., Takizawa, M., Traunmuller, R. (eds.) 15th International Conference on Database and Expert Systems Applications, pp. 13–23. IEEE Computer Society, Zaragoza (2004) 22. Albani, A., Dietz, J.L.G., Zaha, J.: Identifying Business Components on the basis of an Enterprise Ontology. In: Interoperability of Enterprise Software and Applications, pp. 335–347. Springer, Heidelberg (2006) 23. Bachman, C.W.: The role data model approach to data structures. In: Deen, S.M., Hammersley, P. (eds.) International Conference on Databases, pp. 1–18. Heyden & Son (1980) 24. Kristensen, B.: Object-Oriented Modeling with Roles. In: 2nd International Conference on Object-Oriented Information Systems (1995) 25. Reenskaug, T., Wold, P., Lehn, O.: Working With Objects: The OOram Software Engineering Method. Manning Publication Co., Greenwhich (1996) 26. Steimann, F.: On the representation of roles in object-oriented and conceptual modelling. Data & Knowledge Engineering 35, 83–106 (2000) 27. Riehle, D.: Framework Design: A Role Modeling Approach. PhD Zurich, Switzerland: Swiss Federal Institute of Technology, p. 229 (2000) 28. Ould, M.: Business Processes: Modeling and analysis for re-engineering and improvement. John Wiley & Sons, Chichester (1995) 29. Krogstie, J., Carlsen, S., Consulting, A., Chicago, I.L.: An integrated modelling approach for process support. In: 30th Hawaii International Conference onSystem Sciences, HICSS 1997, vol. 2 (1997)

176

A. Caetano and J. Tribolet

30. Wegmann, A.: On the systemic enterprise architecture methodology. In: International Conference on Enterprise Information Systems (ICEIS 2003), Angers, France (2003) 31. Lê, L.-S., Wegmann, A.: SeamCAD: Object-Oriented Modeling Tool for Hierarchical Systems in Enterprise Architecture. In: 39th Hawaii International Conference on System Sciences, Hawaii, USA (2006) 32. Zacarias, M., Caetano, A., Magalhães, R., Pinto, H.S., Tribolet, J.: Towards Organizational Self-Awareness. In: Rittgen, P. (ed.) Ontologies for Business Interactions. Idea Group Inc. (2007) 33. Caetano, A., Rito Silva, A., Tribolet, J.: A Role-Based Enterprise Architecture Framework. In: 24th Annual ACM Symposium on Applied Computing, ACM SAC 2009, Hawaii, USA (2009) 34. Zacarias, M., Magalhães, R., Pinto, H.S., Tribolet, J.: An agent-centric and ‘context-aware’ perspective for the alignment between individuals and organizations. Information Systems (March 14, 2009) 35. Sowa, J.: Conceptual Structures: Information Processing in Mind and Machine. AddisonWesley, New York (1984) 36. Guarino, N., Carrara, M., Giaretta, P.: An Ontology of Meta-Level Categories. In: Principles of Knowledge Representation and Reasoning: Proceedings of the Fourth International Conference (KR 1994), pp. 270–280. Morgan Kaufmann, San Mateo (1994) 37. Carlsen, S.: Comprehensible Business Process Models for Process Improvement and Process Support. In: 8th International Conference, CAISE 1996. LNCS, vol. 1080. Springer, Heidelberg (1996) 38. Giaglis, G.M.: A Taxonomy of Business Process Modeling and Information Systems Modeling Techniques. International Journal of Flexible Manufacturing Systems 13, 209– 228 (2001) 39. March, S., Smith, G.: Design and natural science research on information technology. Decision Support Systems 15, 251–266 (1995) 40. Hevner, A.R., March, S.T., Park, J., Ram, S.: Design science in information systems research. MIS Quarterly 28, 75–105 (2004) 41. Braun, C., Wortmann, F., Hafner, M., Winter, R.: Method construction - a core approach to organizational engineering. In: ACM Symposium on Applied Computing, USA, pp. 1295–1299 (2005)

Towards Traceability Metamodel for Business Process Modeling Notation Saulius Pavalkis1,2, Lina Nemuraite1, and Edita Milevičienė2 1

Kaunas University of Technology, Department of Information Systems, Studentu 50−308, LT-51368 Kaunas, Lithuania 2 No Magic Europe, Savanoriu av. 363, LT-49425 Kaunas, Lithuania [email protected], [email protected], [email protected]

Abstract. This paper presents the traceability metamodel for Business Process Model and Notation (BPMN) and its implementation in Cameo Business Modeler plug-in for MagicDraw. There is no public standard traceability metamodel defined for BPMN yet. We present solutions that we have already applied in practice: we improve the traceability of BPMN models by defining derived properties that are calculated by a modeling tool on the fly. In contrast to other existing solutions, this approach does not require additional efforts from users for defining and maintaining traceability, and does not overload projects with redundant information. Using this approach, CASE tool developers are able to supplement their tools with traceability analysis means allowing users to access traceability information easier, to check completeness and correctness of BPMN models, and to analyze the impact of changes. Keywords: traceability, derived properties, BPMN, model consistency, coverage analysis, change impact analysis.

1

Introduction

Today’s software is becoming more and more complex. Modeling takes an important role in the software development because of the ability to raise the level of abstraction from code to models using popular modeling languages such as UML [2], BPMN [1], SysML [3], and others. Models become primary artifacts in software and systems development. They cover all stages of software development from business analysis and requirements definition to implementation, code generation, and testing, as defined by Unified Software Development Process [4]. As a result, the complexity of models is growing, and this leads to increased risk and higher costs of software projects [5]. In this complex context, it becomes crucial to assure safety, reliability, and quality of software and systems granting their integrity, avoiding redundancy, managing development processes and changes. Model traceability can help to reach these goals since it is able to reduce complexity by easing comprehension of design decisions to stakeholders, decision makers, and developers. T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 177–188, 2011. © IFIP International Federation for Information Processing 2011

178

S. Pavalkis, L. Nemuraite, and E. Milevičienė

Traceability is the important aspect of software development for analyzing risks and costs during change management. On the other hand, BPMN (we are concerning the second BPMN version, BPMN 2) is one of the most popular standards for business process modeling. Many modeling tools support BPMN diagrams [6]. However, traceability of important elements of BPMN models is not assured. In our viewpoint, the traceability information should be created, updated and visualized in such a way that it would not cause more problems than advantages received. It should not unpredictably increase the overhead and costs of the project. This is especially important for business users (i.e. main users of BPMN) working with visual representation of models. Traceability information should be presented in a clear and comprehensive way in order to be understood and accepted by business process modelers. The traceability of BPMN models can be improved by using our proposed derived property approach. The core of the approach is an extension of a metamodel of the problematic modeling language with additional properties that can be calculated by a modeling tool on the fly. In contrast to other existing solutions, this approach does not require users defining and maintaining traceability relations in their projects, and does not overload their projects with traceability information. Using this approach, CASE tool developers are able to supplement their tools with traceability analysis means allowing users to access traceability information easier, to check completeness and correctness of BPMN models, and to analyze impact of changes. The rest of the paper is structured as follows. Section 2 analyzes related works. Section 3 presents the traceability metamodel for BPMN and its implementation in Cameo Business Modeler tool using custom derived properties dedicated for traceability. Finally, section 4 presents conclusions and future works.

2

Traceability Concepts and Related Works

In the IEEE Standard Glossary of Software Engineering Terminology [7], the traceability is defined as “The degree to which a relationship can be established between two or more products of the development process, especially products having a predecessor–successor or master–subordinate relationship to one another”. In order words, traceability is understood as the ability to identify direct or indirect relations among project elements. There are many other traceability definitions; however, most of them are similar to the presented ones. Traceability is classified in different ways on the base of various aspects. According to [8], there are some fundamental classifications, like forward [9], backwards [10], horizontal [11, 12], and vertical traceability [11]. Ramesh and Edwards [11] define the distinction between horizontal and vertical traceability. Traceability that considers links among artifacts belonging to the same project phase or level of abstraction is defined as horizontal traceability; traceability that links artifacts belonging to different phases or levels is defined by vertical traceability. In this paper, we will focus on horizontal traceability links within BPMN models.

Towards Traceability Metamodel for Business Process Modeling Notation

179

Traceability schema, or metamodel of a particular domain defines what relations between specific model elements are treated as traceability relations, and what semantics they carry. Multiple authors have proposed traceability metamodels [13−16] but no common understanding of a complete traceability metamodel [9] is defined yet. The limitation of these approaches arises because of inflexible types of relationships while needs and practices of organizations are changing. A relevant solution should provide traceability metamodel, which supports customization and extensibility of traceability links giving possibility to define new types of links, artifacts, and transitive relations. Such capabilities and predefined schemas are provided in our derived properties based approach which is already applied to UML and SysML. BPMN role in software development is discussed in multiple sources [17]. However, there is no standard traceability metamodel for BPMN. The lack of traceability in BPMN models causes a number of problems: 1st problem. Resource roles take part in BPMN activities that belong to some process. However, there is no direct relation between resource role and process in BPMN metamodel. This means it is impossible to trace information about processes in which the resource role takes part. For solving this problem, we introduce a traceability rule defining relation between resource role and process, and vice versa. 2nd problem. Identification of business concepts is crucial for business process modeling, and we apply UML class diagram for this purpose [17]. Instances of concepts taking part in workflows of business process models are represented as objects. However, it is not possible to find out in which processes business concepts take part. We propose a traceability rule defining relation between business concept (class or BPMN resource) and processes, in which it takes part, and vice versa. 3rd problem. Participants take part in sending and receiving messages during process execution. In BPMN metamodel, there is no direct relation between participants and messages sent or received by them. Consequently, we introduce a traceability rule defining messages sent and received by participants. One of the most important aspects of traceability analysis tools is their ability to represent results. Winkler et al. emphasizes matrices, cross-references, and graphbased representations as main methods for traceability visualization [9], and summarizes other traceability benefits: prioritizing requirements, estimating change impact, proving system adequateness, understanding the system, supporting design decisions, validating and much more. Cameo Business Modeler covers methods for traceability visualization together with other visualization and analysis means. Also, we provide other capabilities for analyzing models on the base of traceability information: change impact analysis [18], checking consistency and completeness of models.

3

Traceability Metamodel for Business Process Model and Notation

We solve BPMN traceability problems by applying custom derived properties approach which allows to extend UML metamodel for derived properties

180

S. Pavalkis, L. Nemuraite, and E. Milevičienė

specification and enables its customization as a part of MagicDraw DSL engine − a core of Cameo Business Modeler [19]. As the mentioned BPMN traceability problems are caused by absence of direct relations between elements, we use derived properties for creating such relations. Definition and calculation of derived properties are presented in section 3.1 “Derived Properties Framework and Metamodel” (full description may be found in [20]). We define BPMN traceability rule expressions as property chains (Table 1) where column “Rule name” shows the name of the derived property; “Source element” identifies the owner of that derived property; “Target element” corresponds to a value of the derived property. Table 1. BPMN traceability rules solving traceability problems Rule name Source element Target element Resource Role – BPMN Process traceability 1. Taking Part in Resource Role BPMN Process Process 2. Activity BPMN Process Resource Role Resources Business Concept – BPMN Process 3. Taking Part in Class or BPMN BPMN Process Process resource 4. Business concepts BPMN Process Class or BPMN resource Participant – Message 5. Sent Messages Participant

6. Received Messages

3.1

Participant

Message

Message

Description Defines processes, in which activity resource role takes part. Defines resources roles taking part in process activities. Defines processes, in which business concepts takes part. Defines business concepts, which take part in business process. Participants take part in sending and receiving messages during process execution. Property defines messages sent by participant Participants take part in sending and receiving messages during process execution. Property defines messages received by participant via message flow from representing pools.

Derived Properties Framework and Metamodel

In order to be able to define derived properties in modeling environment we extended a modeling language for specifying derived property details. UML and other modeling languages provided by Object Management Group (OMG) have the standard extension mechanism – profiling.

Towards Traceability Metamodel for Business Process Modeling Notation

181

UML extension for derived properties reuses UML properties and MagicDraw DSL engine [19] constructs; it introduces only one stereotype with a single property for specifying derived properties. The stereotype DerivedPropertySpecification extends UML metaclass Property for specification of derived property expression that defines how this property is calculated. UML properties of stereotyped property element are used for specifying a derived property as well: one can define the name, type, multiplicity, isUnique, isOrdered, isDerived, isReadOnly, and body of comment. Stereotyped property with DerivedPropertySpecification stereotype is added into MagicDraw DSL [19] customization class (i.e. class stereotyped as customization). The definition of tag customizationTarget of this class specifies in which element type (UML or extended one) derived property will be created (for more details of specifying derived properties please refer to [20]). The heart of the derived property is the expression according to which it is calculated. Several different types of such expressions are available in Cameo Business Modeler: simple − through UML properties and relationships; more advanced ones use OCL expressions, property chains, scripting languages or Java code. Most popular are simple expressions and property chains. Property chain expression type (i.e. path through metamodel and properties) is used for navigating from a context element to a final linked property for gathering the resulting elements as derived property values. A property chain expression defines navigation path consisting of metaclass/stereotype and property/tag pairs. For example, for derived property “Taking Part in Process” (Fig. 2), the property chain expression is ResourceRole.opposite(BPMNActivity.resources).owner [20]. For derived property “Received Messages” (Fig. 8) property chain expression is Participant. Opposite (ActivityPartition.represents). opposite (DirectedRelationship.source). messageRef [20]. Derived property expressions could be written in OCL. For example, derived property “Taking Part in Process” expression could by written in OCL as context ResourceRole::takingPartInProcess:BPMNProcess derive: self.BPMNActivity.owner, and “Received Messages” could be written in OCL as context Participant::receivedMessage:Bag(Message) derive:self.activityPartition.directedRelationship select(n| n.oclIsKindOf(MessageFlow)).messageRef However, our current traceability implementation in MagicDraw allows users to specify property chains in a simpler way by choosing required elements and properties in dialog. OCL expressions could be applied for more complex situations.

182

3.2

S. Pavalkis, L. Nemuraite, and E. Milevičienė

Traceability Rules

Resource Roles taking part in BPMN activities (tasks and subprocesses) are shown in Fig. 1. BPMN standard property − Resources shows all resource roles taking part in a particular activity.

Fig. 1. BPMN process diagram representing registration for open training class

For assuring Resource Role – Process traceability, we introduce bidirectional traceability relation between resource role and BPMN process (Fig. 2). In result, we can see all resources, participating in activities of a particular BPMN process, in the specification of that process (Fig. 3).

Fig. 2. BPMN metamodel extension for traceability relation between resource role and BPMN process

Towards Traceability Metamodel for Business Process Modeling Notation

183

Fig. 3. Derived property “Activity Resources” in the BPMN process specification

Definition of business processes can start from identifying business concepts. For this purpose, we can use UML class diagram [17]. Business concepts (classes or BPMN resources) taking part in the workflows of business processes are represented as data objects (Fig. 4). However, there is no possibility to trace from business concepts to processes where instances of concepts (data objects) are used.

Fig. 4. BPMN diagram for Order Handling Process

Metamodel for solving Business Concept − BPMN Process traceability problem is presented in Fig. 5. This metamodel introduces a relation between business concept (class or BPMN resource) and process in which it takes part. This extension allows to show all business concepts, taking part in BPMN process, in MagicDraw Dependency Matrix (Fig. 6). Resource which is not used as business concept will still have traceability property, but it will not have a value pointing to BPMN process.

184

S. Pavalkis, L. Nemuraite, and E. Milevičienė

Fig. 5. Parts of BPMN and UML metamodels extended with traceability relation between Resource and BPMN process

Fig. 6. Visualization of traceability property “Taking Part in Process” with Dependency Matrix

Similarly, BPMN collaboration diagram is the only place where participants and messages sent or received by them can be seen (Fig. 7). There are no direct relation between participant and message in BPMN metamodel. For assuring this kind of traceability, we introduce two traceability relations between Participant and Message (Fig. 8). These relations allow to show messages of a particular participant in its specification (Fig. 9).

Towards Traceability Metamodel for Business Process Modeling Notation

185

Fig. 7. BPMN Collaboration diagram showing messages sent between participants

Fig. 8. Parts of BPMN and UML metamodels extended with traceability relation between participants and messages sent or received by the participant

186

S. Pavalkis, L. Nemuraite, and E. Milevičienė

Fig. 9. Participant specification extended with derived traceability properties – Sent Messages and Received Messages

3.3

Visualization and Analysis of Traceability Rules

Once derived properties are specified, they appear in specification dialog of the corresponding elements and other places in the same way as regular BPMN properties. Now, by visualizing and analyzing traceability information, we can discover related elements, which will be impacted by model changes (i.e. we can perform impact analysis). Impact analysis is performed by discovering impacted parts – the ones related with traceability relations. The following paragraph overviews methods for discovery of impacted parts. Also, we can validate model consistency by performing coverage analysis for discovering whether all requirements are satisfied by design and verified with test cases, or not. Transitive traceability can be visualized by Relation Map − a graph based visualization that allows review and analysis of multilevel relations. Dependency Matrix represents traceability relations between requirements and design. Traceability properties also can be visualized on diagram using standard MagicDraw mechanism for displaying property values in notes. Traceability property groups can be seen in the contextual menu of an element. Generic table is an easy way for performing coverage analysis e.g. empty cells in the rows indicate a lack of consistency in the model. Coverage analysis report can be generated using documentation generation capability. All the derived properties, together with custom BPMN properties, can be accessed when creating user specific report templates. Finally, you can check completeness of traceability and validate non-existence of cyclic relationships by using CBM validation feature. Predefined validation suite for traceability checks model for empty traceability properties and elements involved in both forward and backward traceability relations with another elements.

Towards Traceability Metamodel for Business Process Modeling Notation

4 1.

2.

3.

4.

5.

6.

187

Conclusions and Future Works Analysis of traceability in current BPMN 2 models had shown the lack of traceability between BPMN concepts, from which we first had taken into consideration processes and resource roles, BPMN processes and business concepts, participants and messages as most required ones. As the solution for these problems, we propose traceability metamodel based on UML derived properties. Our proposed derived property approach is already tried in practice for UML and SySML, and could be adopted by other developers in other tools and for other languages as currently we have done for BPMN. In contrast to other proposals, derived property based traceability framework supports customization and extensibility of traceability links giving possibility to define new types of links, artifacts, and transitive relations. Also, it has advantages with significant decrease of overhead as derived properties are automatically calculated by a modeling tool and dynamically updated according to changes in models. No manual work is required to specify traceability information − it is created and updated fully automatically. Implementation of traceability metamodel in Cameo Business Modeler (CBM) allows reusing existing CBM means as dependency matrices, report templates, and validation rules for traceability information analysis, visualization and navigation. Proposed BPMN 2 traceability metamodel and rules provide information about dependencies between BPMN processes and resource roles, processes and business concepts, participants and messages. This allows validating BPMN 2 models for correctness and completeness of these aspects, and analyzing impact of changes. In our future work, we will concentrate on extending the traceability metamodel for BPMN 2 as well as defining specific traceability metamodels for modeling databases and enterprise architectures.

Acknowledgement. The authors would like to thank No Magic, Inc, especially the MagicDraw and Cameo Business Modeler product teams for comprehensive support.

References 1. OMG: OMG Unified Modeling Language (OMG UML), Superstructure, Version 2.3. OMG Document Number: formal/2010-05-05 (2010) 2. OMG: Business Process Model and Notation (BPMN), Version 2.0. OMG Document Number: formal/2011-01-03 (2010) 3. OMG: OMG Systems Modeling Language (OMG SysML), Version 1.2. OMG Document Number: formal/2010-06-01 (2010) 4. Jacobson, I., Booch, G., Rumbaugh, J.: The Unified Software Development Process. Addison-Wesley Professional, Boston (1999) 5. Biffl, S., Aurum, A., Boehm, B., Erdogmus, H., Grünbacher, P.: Value-based software engineering. Springer-Verlag New York, Inc., Secaucus (2005)

188

S. Pavalkis, L. Nemuraite, and E. Milevičienė

6. Allweyer, T.: BPMN 2.0 – Introduction to the Standard for Business Process Modeling. Books on Demand GmbH (2010) 7. IEEE Standards Board.: IEEE Standard Glossary of Software Engineering Terminology, IEEE Std 610.12-1990. IEEE Press, Piscataway (1990) 8. Aizenbud-Reshef, N., Nolan, B.T., Rubin, J., Shaham-Gafni, Y.: Model traceability. IBM Systems Journal 45(3), 515–526 (2006) 9. Winkler, S., Pilgrim, J.: A survey of traceability in requirements engineering and modeldriven development. Software and Systems Modeling 9(4), 529–565 (2010) 10. IEEE Standards Board: IEEE Guide to Software Requirements Specification, ANSI/IEEE Std 830-1984. IEEE Press, Piscataway (1984) 11. Ramesh, B., Edwards, M.: Issues in the development of a requirements traceability model. In: Proceedings of the IEEE International Symposium on Requirements Engineering, pp. 256–259. IEEE Computer Society, New York (1993) 12. Briand, L.C., Labiche, Y., Yue, T.: Automated traceability analysis for UML model refinements. Information and Software Technology 51, 512–527 (2009) 13. Ramesh, B., Jarke, M.: Toward reference models for requirements traceability. IEEE Transactions on Software Engineering 27(1), 58–93 (2001) 14. Pinheiro, F.A.C., Goguen, J.A.: An Object-Oriented Tool for Tracing Requirements. IEEE Software 13(2), 52–64 (1996) 15. Dick, J.: Rich Traceability. In: Proceedings of the 1st International Workshop on Traceability in Emerging Forms of Software Engineering, Edinburgh, Scotland, pp. 18–23 (2002) 16. Letelier, P.: A Framework for Requirements Traceability in UML-Based Projects. In: Proceedings of the 1st International Workshop on Traceability in Emerging Forms of Software Engineering, Edinburgh, Scotland, pp. 30–41 (2002) 17. Silingas, D., Butleris, R.: Towards implementing a framework for modeling software requirements in MagicDraw UML. Information Technology and Control 38(2), 153–164 (2009) 18. Crnković, I., Asklund, U., Persson-Dahlqvist, A.: Implementing and Integrating Product Data Management and Software Configuration Management. Artech House, London (2003) 19. Silingas, D., Vitiutinas, R., Armonas, A., Nemuraite, L.: Domain-specific modeling environment based on UML profiles. In: Information Technologies 2009: Proceedings of the 15th International Conference on Information and Software Technologies, IT 2009, Kaunas, Lithuania, April 23-24, pp. 167–177. Kaunas University of Technology, Technologija (2009) 20. UML Profiling and DSL User Guide, https://secure.nomagic.com/files/manuals/ UML%20Profiling%20and%20DSL%20UserGuide.pdf

The Connected Home: From Market Barriers to Business Model Solutions Malgorzata Ciesielska and Feng Li RCUK Digital Economy Research Hub on Social Inclusion through the Digital Economy (SiDE), Newcastle University Business School, NE1 7RU United Kingdom {m.ciesielska,feng.li}@nlc.ac.uk

Abstract. The market for the connected homes is still immature, and there is a considerable confusion about the possible ways of their market development. Basing on a systematic review of previous studies, we analyse nine main reasons why the connected home market has so far failed to take off. We conclude that consumer misperceptions, lack of knowledge as well as the closed system architecture by most providers are the key barriers to their largescale development. We suggest that overcoming those barriers is possible by referring them to business model building blocks. Keywords: connected home, smart home, market barriers, business model, interoperability, communication protocols standardisation.

1

Introduction

A connected home is a high-tech residential setting, which incorporates ICT to offer particular services to the residents for the smart assisted living. Smart and connected homes, alongside the notion of assisted living in particular, have gained a lot of attention in the last few years. This is the result of several trends: ubicomp revolution of 90s [1-3], widespread proliferation and fast adaptation of mobile technology and a steady growth rate of households with Internet access all over the world at the similar phase (worldbank.org data) mixed with ageing society- related problems, including the growing health and social-care costs [4-6]. But, the fully connected home has not yet fully emerged and we still don’t recognise clear business models of how to bring connected homes and assistive technologies to the mass market. This paper presents the results from our analysis of yet not overcome connected home market development barriers. To identify them we applied the systematic literature review method [7-8] within Thompson’s ISI Web of Knowledge database in all available years (1970-2010). As an outcome, we came up with 39 papers that clearly referred to the market barriers and business challenges for connected home’s solutions. The contribution of this paper is twofold. Not only we indicate the recurring problems of the connected home market but in we clearly refer them to the areas of key business activities, in which they should be addressed. In order to do so we T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 189–199, 2011. © IFIP International Federation for Information Processing 2011

190

M. Ciesielska and F. Li

employ the business model perspective. Following [9-10]: we concentrate on three business model components: Product; Infrastructure; and Customers. To be sustainable, the business has to be financially stable and credible in front of its stakeholders. This sustainability is reach by building positive customer relationship, attention to consumer needs and keeping them informed about the company’s actions. At the same time the business infrastructure, its internal and external networks, has to be well thought and directed to match the product/service value proposition. Value proposition, especially the outcomes of product and services development and commercialisation, is a key to revenue generation.

2

Market Barriers

As far as at the beginning of the new millennium two industry reports [11-12] indicated that markets, technologies and supply chains of connected homes are immature and that the consumers are not only sceptical but also ignorant of technological improvements and that building industry should provide generic infrastructure. Not only the demand seems to be inactive, but what is worse the demand is unconscious, unaware of what is available on the market [13]. In addition to that, although end-users express worries about privacy issues, they haven’t been so far addressed properly [14]. Moreover, the costs of switching to such technologies appeared to be to high, both in financial terms and private time and habits. In recent years a marketing strategy for the connected home concept changed back to offering single devices enabled to be connected and integrated in larger systems when necessary and configured according to particular buyer’s needs. Not a wonder that at the same time construction and property industry was little enthusiastic about creating a market for a fully automated, intelligent house [13]. At this stage of the connected home market creation, the business perspective is needed. The customers have to see a clear value creation chain, which ensure the services and advantages of the smart technologies to be sustainable and justify investments in it. Suppliers, although are more successful with selling separate devices / appliances rather than offering integrated systems, they will soon meet other demand barriers. Early adopters may be happy with trying out technical novelties, but larger population expects fully usable products, not a promise of possible functions. In the following sections we will summarise the main market-related problems of connected homes, in particular to the user-related issues, technology & supply-related issues, costs, and finally business-related issues including strategy, marketing and services. 2.1

Users Perceptions

The connected home is often seen through analogies to the futuristic, science-fiction stories and movies. The wider public is not informed about possibilities or advantages of the connected homes and often links them to visions taken from pop culture: “The Jetsons” style of living [15] or “Start Trek” [16]. On the other hand, there seem to be the lack of proper information campaign about the connected home benefits, like the ones for energy-saving, security, safety, convenience and enhancing communication

The Connected Home: From Market Barriers to Business Model Solutions

191

[17] and there are still a number of particular acceptance barriers, which need to be specifically addressed, like: price stability, lack of information, compatibility and possibilities of upgrades, reliability, servicing costs, and complexity in use [4]. 2.2

Poor Understanding of User Needs

According to Paul Liao - the president of Panasonic technologies - in the past 30 years progress in technology led to many new products, but its success depends on consumers’ expectations, which are steadily growing [18]. In other words customers always expect even further improved performance, decreased costs, miniaturisation, etc. The fact that the technology-oriented approach had failed to provide significant understanding of domestic consumers’ expectations was noticed first in 1995 [15]. But similar critics of the innovation industrially driven can be found also in more recent studies [4, 11, 19-20]. It seems that ubicom revolution hasn’t managed to produce true user-oriented value in the connected home area. 2.3

Data Security and Privacy Issues

There are several security and privacy issue that haven’t so far been resolved. This includes authentication and authorisation [21], a need for low-cost, high-quality and excellent security services [22], dependability problems of reliability, availability, security, and manageability [23] or a need for identifying of security mechanisms needed to ensure privacy [24-25]. Once the monitoring systems are installed, the technical possibilities of monitoring are always available. “It could be argued that the ethical pitfalls in monitoring people’s movements within their home, or recording the frequency of using the toilet or regularity of eating, counteract the advantages of various smart home technologies for providing independence” (14, p. 17). Therefore, privacy should be a central design issue in its own right with a particular focus in the design areas of capture of data, construction, accessibility and purpose. The key problems lay in disembodiment of the collected data from the context and dissociation from one's actions. The solution is better control and feedback [14]. 2.4

Lack of Effective Marketing Message

There is very little PR or marketing strategies aiming at informing the potential customers about the possibilities and advantages of the connected home technologies as well as to dispel doubts about safety, security and other technology-related issues. The problems of lack of the proper marketing communication were raised from the beginning of the 90s [26] and this situation continued over the next two decades [12, 13, 19, 27]. 2.5

Lack of Installation, Maintenance Services and Skills

The large reports on connected homes at the turn of the 1990s indicated the obvious lack of skills among the builders who were supposed to fit the smart technologies in the buildings [11-12, 29]. Similarly [30] and [20] emphasise that still most potential

192

M. Ciesielska and F. Li

users don’t have enough technical skills to install and maintain their connected home systems on their own. Without infrastructure solutions, it may be very difficult to provide smart home installations outside the DIY niche. 2.6

The Costs of Devices and Installations

The customers are interested in connected home solutions but are unwilling to pay the current prices [17, 31-33]. From a house developer perspective, the full range of intelligent technologies would add min +5% to the new home costs [17] or an average about £1400 per room [11]. Because the relatively high initial investment form the consumer is still necessary [12, 34], those costs make the connected home targeted at middle and upper income groups while it stays unreachable for others who could possible benefit form it, like the disabled or the elderly [4]. 2.7

The Old Housing Stocks and Retro-adaptation Problems

The large old house stock, in particular in Europe, is another obstacle to the consumer up-take of connected home technology [34]. It is calculated that networking an old building is more expensive than doing it during the new construction works [4]. Therefore, the full retro-fit to existing building is considered improbable on the wide scale due to financial and disruption costs [12,19]. 2.8

The Supply: Pessimistic about the Market, Poorly Developed Distribution Channels

The connected home undertaking requires involvement of various specialists originating from retail, transport, services, installation, house-building and software industries and their collaboration has to be properly managed. That is why we believe that the future of the connected home market will depend on the system integrators [4]. Possibly, the slow development of the connected home market reflects the lack of the genuine successful business case in this area. It seems that not only building industry is caught in the concept of constantly futuristic market but also consumer electronic industry considers it as undefined future opportunity [36]. 2.9

Lack of Common Standards and Therefore Difficulties in Integrating Different Systems

Huang et al. [37, p. 619] “hypothesize that the utility of input appliances will be greatly increased if they too were ‘infrastructure enabled”. But they still aren’t. The problem is in the lack of common standards, which concerns many issues primarily related to: communication protocols and control [4, 11, 19, 26, 28, 29, 38, 40-46] but also standardisation of testing and diagnostic tools [24]. As early as 1987 the New Scientist reported that “Philips, Thorn-EMI, Thompson. Siemens, GEC, Mullard and Electrolux decided that their next big market would be the “intelligent home” [38]. Ten years after, Barlow & Gann [4] enumerate several “standards” existing on the market in 1998 in the US NHBA: Smart House and EIA:

The Connected Home: From Market Barriers to Business Model Solutions

193

CEBus, in Europe LON Users Club, EHS Association, EIB Association, BatiBUS Club, in Japan: TRON house and EIAJ: HBS. The 90s are described as the time of “battle for dominance” over the concept of the connected home and in particular the standardisation attempts [46, 47]. While the connected home moved towards consumer electronic market and revealed its multi-paradigm setting and open character, the actors kept referring to it in the system paradigm with tight coordination of innovation processes. The trials of integrating some of those standards were to a large extend unsuccessful and still the easiest and fully open ones X10, ZigBee and XBee are also the most popular for connected home DIY. Currently the most known and widely applied consumer electronic communication standard is UPnP with DLNA certificate. The Digital Living Network Alliance consists of over 200 companies who seek to create new products that are compatible by using open standards and widely available industry specifications. The promote members are companies like Access, AT&T, AWOX, Broadcom, CableLabs, CISCO, Direct TV, DOLBY, DTS, Ericsson, HP, Huawei, Intel, LG Electronics, Microsoft, Motorola, Nokia, Panasonic, Pioneer, Qualcomm, Rovi, Samsung, SHARP, SONY, Technicolor, Toshiba, and Verizon. Although UPnP/DLNA is developed by a group of companies, it doesn’t mean that their products are all compatible. On contrary, the standard has various implementations, which creates burdens for it to work properly. It is enough to follow any Internet forum describing the issues concerning UPnP problems or look at the projects like Rygel/GUPnP to realise that different UPnP/DLNA hardware is not interoperable. Some companies, like Microsoft seem to have a purposeful strategy of making their devices inoperable with some of their competitors. At the end of the day, what was supposed to be automatic and easy, is in fact just an unfulfilled promise. Moreover, the industry consortium restricts the type of formats supported by UPnP/DVLA, which results in the often error messages like “Unsupported format” or “Data is corrupted” which makes this standard even less useful. Not surprising that many involved companied do not even inform their customers that their products are adopting UPnP/DVLA standards [30]. The contradictory, sand-alone strategy is represented by Apple - a notable nonmember of Digital Living Network Alliance with its own technology. Apple products are known to operate well among each other and with most other manufacturers’ devices, for instance printers via Bonjour protocol. At the same time Apple standards are not fully open, and probably services like video transmission will be still available just in one direction, from others devices to Apple devices. On contrary to UPnP/DLNA, Apple’s devices do what the manufacturer promise to.

3

Market Barriers and Business Model Challenges

The reviewed literature mentioning the key barriers of the connected home market can be classified into nine groups of problems (Table 1 below). But our aim was not only to create the list of existing market barriers for development of connected homes, but also to relate them to the key elements of a business model. Business model is an analytical way of looking at entrepreneurial endeavour by focusing on the value propositions, ways of generating profits and necessary facilities and arrangements.

M. Ciesielska and F. Li

1987

New Scientist [38]

1989

Hanover [29] Gann [28]

1991

Haddon [26] Weiser [1]

1992 1993 1995

Lutolf [43] Yang & Manikopoulos [39] Bellotti & Selen [14] Cawson et al. [35] Haddon [15] Barlow & Gann [4]

1998

Badami & Chbat [31] Tsai et al. [22] vanBerlo & Fellbaum [44]

1999

Charles [40] Fellbaum & Hampicke [41] Gann et al. [11] Anderson [16] Al-Muhtadi et al. [21] Maglio et al. [48] Huang et al. [37]

2000

Kiciman & Fox [45] Pragnell et al. [12] Rosenthal & Stanford [24] Spohrer & Stein [32] Wang et al. [23]

2001

Petersen et al. [17]

2003

Barlow & Venables [13]

Aldrich [34] Tronnier [42] 2004

David [27]

Old housing stock

Poor & pessimistic supply

User perceptions

marketing

Lack of skills

Security & privacy issues

users' needs

The costs

Table 1. Key market-development barriers Lack of common standards

194

The Connected Home: From Market Barriers to Business Model Solutions

195

Table 1. (Continued) Liao [18] Beniaminy [33] 2007

Bierhoff et al. [19] Coughlin et al. [20] Tamura et al. [25]

3.1

2008

Peine [46]

2009

Peine [47]

2010

Bidmead [30]

Customer Relationship Management

First, there are several serious problems with customer relationship management. On the one hand users tend to have futuristic, partly unrealistic connotations. On the other hand they lack of knowledge about the available technology and there are issues about the proper product classification and guarantee. For instance, within the assisted living technologies people seem to look for either products like medical devices where their reliability is guaranteed by the law. All of those problems should be addressed by the proper CRM, which so far was mostly neglected by the industry. 3.2

Connected Home Innovation and Commercialisation

Second, the price and affordability for different market segments, including nonfinancial costs of panopticon independency and stigma stays unsolved. For instance, the elderly are looking for more generic products that are not designed for “old” or sick” but just for “younger or more healthy older adults seeking to simply age-inplace.” [20 p.1813]. Also matters of security and privacy are not clear to the end-users. Although from the industry point of view often high prices for the connected home systems are justified by the cost of innovation, there is little explanation of the value to the end users. But the question of price is more complicated and refers to the other important issue who will pay for the connected home technology, especially assisted living one: the elderly or sick themselves, the families, the government? Careful stakeholder analysis would help in creating sustainable business models able to commercialise successfully smart technologies for connected home. 3.3

Infrastructure Management

Finally, the pessimism on the supply side with high costs of house retro-adaptation and lack of technical skills makes infrastructure management difficult. But this indicates on more major choice, which lies across the sections of Customer Relationship Management, product/service innovation and commercialisation and infrastructure management and therefore can be understood as a main connected home business model variable: the lack of common standards. Therefore on the

196

M. Ciesielska and F. Li

supply side, there is an important choice to be made between open and closed business models with in results implicate different modes of cooperation and different phases of market development. So far, neither quasi-open strategy of Microsoft nor the closed strategy of Apple was able to help gaining significant competitive advantage.

4

Conclusions

The most commonly mentioned market barriers for connected homes are: user perceptions, poor understanding of users needs, security issues, lack of effective marketing, installation and maintenance skills, costs, old housing stock, pessimism in the industry, and lack of common communication standards and data formats. However, they seem to be considered as an explanation why the technology didn’t reach the mass-market rather than the set of problems that must be solved. Therefore, our analysis shows that the market-development issues were raised for many years, and still fail to be properly addressed. One of the reasons for this situation is lack of business-oriented framework, and considerations in terms of models of value creation. We predict that the future success of the connected home approach will depend critically on the widespread adoption of the “webs” [49] and platform strategy [50] based on common standard and increasing returns. But this has wider consequences, as without the common standards; the mass market is unlikely to take off. However, the interoperability contradicts the monopolistic and/or lock-in strategies pursued by some of the significant players from the building or the home electronics industries. Many of them are currently adopting a wait-and-see strategy, risking that their competitors may come up with new killer applications or new modes of cooperation [51]. Acknowledgement. This project is part of Social inclusion through the digital economy hub financed by Research Council.

References 1. Weiser, M.: The Computer of 21st Century. Scientific American 265(3), 94–104 (1991) 2. Weiser, M.: Some Computer-Science Issues In Ubiquitous Computing. Communications of the ACM 36, 75–84 (1993a) 3. Weiser, M.: Ubiquitous Computing. Computer 26, 71–72 (1993b) 4. Barlow, J., Gann, D.: A changing sense of place: are integrated IT systems reshaping the home? Electronic Working Paper Series, Science Policy Research Unit (SPRU). University of Sussex, Paper No. 18 (1998) 5. Becker, S.A., Webbe, F.M.: The Potential of Hand-held Assistive Technology to ImproveSafety for Elder Adults Aging in Place. In: Henriksen, K., Battles, J.B., Keyes, M.A., et al. (eds.) Advances in Patient Safety: New Directions and Alternative Approaches. Technology and Medication Safety, vol. 4, Agency for Healthcare Research and Quality (US), Rockville (2008), http://www.ncbi.nlm.nih.gov/books/NBK43759/ #advances-becker_3.s1

The Connected Home: From Market Barriers to Business Model Solutions

197

6. Mynatt, E.D., Essa, I., Rogers, W.: Increasing the opportunities for aging in place. In: Cuu 2000 Conference Proceedings, pp. 65–71 (2000) 7. Tranfiled, D., Denyer, D., Smart, P.: Towards a methodology for developingevidenceinformed management knowledge by means of systematic review. British Journal of Management 14(3), 207–222 (2003) 8. Lee, R.: Social capital and business and management: Setting a research agenda. International Journal of Management Reviews 11(3), 247–273 (2009) 9. Osterwalder, A., Pigneur, Y.: An e-Business Model Ontology for Modeling e-Business. In: 15th Bled Electronic Commerce Conference e-Reality: Constructing the e-Economy, Bled, Slovenia, June 17-19 (2002) 10. Li, F.: What Is E-Business? How the Internet Transforms Organizations. Blackwell Publishing (2007) 11. Gann, D., Barlow, J., Venables, T.: Digital futures: Making Homes Smarter. The Joseph Rowntree Foundation, York (1999) 12. Pragnell, M., Spence, L., Moore, R.: The Market Potential for Smart Homes. Joseph Rowntree Foundation, York (2000) 13. Barlow, J., Venables, T.: Smart home, Dumb Suppliers? The future of Smart Home Markets. In: Harper, R. (ed.) Inside the Smart Home, pp. 247–262. Springer, London (2003) 14. Bellotti, V., Sellen, A.: Design for Privacy in Ubiquitous Computing Environments. In: Proceedings of the Third European Conference on Computer-Supported Cooperative Work (ECSCW 1993), pp. 77–92 (1993) 15. Haddon, L.G.: Home automation: Research issues. Paper Presented at the Second EMTEL Workshop: The European Telecom User, Amsterdam, November 10-11 (1995) 16. Anderson, M.K.: Becoming Your Own Hospital. Wired news (2000), http://www.wired.com/science/discoveries/news/2000/11/40120 (retrieved September 1, 2010) 17. Petersen, T., Williams, P., Mills, A.: Analysis of the value of home automation systems. Facilities 19(13/14), 522–530 (2001) 18. Liao, P.: Taking ideas from life to create life. In: de Bernabe y Varela, F.G. (ed.) Connected Homes.Thought Leaders. Essays from Innovators, Cisco Systems, pp. 50–55 (2004) 19. Bierhoff, I., Van Berlo, A., Abascal, J., Allen, B., Civit, A., Fellbaum, K., Kemppainen, E., Bitterman, N., Freitas, D., Kristiansson, K.: Smart home environment. In: Roe, P.R.W. (ed.) Towards an Inclusive Future: Impact and Wider Potential of Information and Communication Technologies, pp. 110–156 (2007), http://www.tiresias.org/cost219ter/inclusive_future/ 20. Coughlin, J.F., D’Ambrosio, L.A., Reimer, B., Pratt, M.R.: Older adult perceptions of smart home technologies: Implications for research, policy & market innovations in healthcare. In: 2007 Annual International Conference of the IEEE Engineering in Medicine and Biology Society, vol. 1-16, pp. 1810–1815 (2007) 21. Al-Muhtadi, J., Anand, M., Mickunas, M.D., Campbell, R.: Secure Smart Homes using Jini and UIUCSESAME. In: Proceedings of the 16th Annual Computer Security Applications Conference, pp. 77–85 (2000) 22. Tsai, S.M., Yang, P.C., Wu, S.S., Sun, S.S.: A Service of Home Security System on Intelligent Network. IEEE Transactions on Consumer Electronics 44, 1360–1366 (1998) 23. Wang, Y.M., Russell, W., Arora, A., Xu, J., Jagannathan, R.K.: Towards dependable home networking: An experience report. In: Proceedings of International Conference on Dependable Systems and Networks, DSN 2000, pp. 43–48 (2000)

198

M. Ciesielska and F. Li

24. Rosenthal, L., Stanford, V.: NIST Smart Space: Pervasive computing initiative. In: Proceedings of IEEE 9th International Workshops on Enabling Technologies: Infrastructure for Collaborative Enterprises, pp. 6–11 (2000) 25. Tamura, T., Kawarada, A., Nambu, M., Tsukada, A., Sasaki, K., Yamakoshi, K.: EHealthcare at an Experimental Welfare Techno House. Japan Open Med. Inform. J. 1, 1–7 (2007) 26. Haddon, L.G.: The Intelligent Home. Investment, Procurement and Performancein Construction, 340–347 (1991) 27. David, S.: Consumers take the lead. In: de Bernabe y Varela, F.G. (ed.) Connected Homes. Thought Leaders. Essays from Innovators” Cisco Systems, pp. 44–49 (2004) 28. Gann, M.D.: The Diffusion of Intelligent Building Technologies. In: Venmore-Rowland, P., Brandon, P., Mole, T. (eds.) Investment, Procurement and Performance in Construction, pp. 328–334 (1991) 29. Hanover, G.: Networking the Intelligent Home. IEEE Spectrum 26, 48–49 (1989) 30. Bidmead, C.: WTF is... DLNA? Hardware (2010), http://www.reghardware.com/2010/10/18/wtf_is_dlna/ (accessed October 18, 2010) 31. Badami, V.V., Chbat, N.W.: Home appliances get smart. IEEE Spectrum 35, 36–43 (1998) 32. Spohrer, J., Stein, M.: User experience in the pervasive computing age. IEEE Multimedia 7, 12–17 (2000) 33. Beniaminy, I.: Smart Scheduling For Smart Meters professional technology & business smart appliance everyone. Pipeline & Gas Journal, 32–34 (September 2007) 34. Aldrich, F.K.: Smart Homes: Past, Present and Future. In: Harper, R. (ed.) Inside the Smart Home, pp. 17–40. Springer, London (2003) 35. Cawson, A., Haddon, L., Miles, I.: The Shape of Things to Consume, ch. 3. Home automation (1995), http://www2.lse.ac.uk/medialse/whosWho/AcademicStaff/ LeslieHaddon/ShapeofThings.pdf 36. Craumer, M.: The sputtering R&D machine. Harvard Business Review 80, 25+ (2002) 37. Huang, A.C., Ling, B.C., Barton, J.J., Fox, A.: Running the Web backwards:appliance data services. Computer Networks-the International Journal of Computer and Telecommunications Networking 33, 619–631 (2000) 38. European Companies Set the Standard for Smart Homes. New Scientist 113, 31 (1987) 39. Yang, J., Manikopoulos, C.N.: Router Connected Physical Media Innetworking the Intelligent Home. IEEE Transactions on Consumer Electronics 38, 30–35 (1992) 40. Charles, J.: Ubiquitous computing uncorked. IEEE Software 16, 97–99 (1999) 41. Fellbaum, K., Hampicke, M.: Integration of smart home components into existingresidences. Assistive Technology on the Threshold of the New Millennium 6, 497– 501 (1999) 42. Tronnier, H.: CENELEC Work Shop - SMart House, Final report, Euronom, Kronberg (2003), http://www.cenelec.eu/Cenelec/Technical+work/ CENELEC+Workshop/Ongoing+Workshops/Smart+House.htm 43. Lutolf, R.: Smart Home Concept and the Integration of Energy Meters into a Home Based System. In: Seventh International Conference on Metering Apparatus and Tariffs for Electricity Supply, vol. 367, pp. 277–278 (1992) 44. Van Berlo, A., Fellbaum, K.: Smart home technology: Useful applications for older people. Assistive Technology on the Threshold of the New Millennium 6, 508–512 (1999)

The Connected Home: From Market Barriers to Business Model Solutions

199

45. Kiciman, E., Fox, A.: Using Dynamic Mediation to Integrate COTS Entities in a Ubiquitous Computing Environment. In: Thomas, P., Gellersen, H.-W. (eds.) HUC 2000. LNCS, vol. 1927, pp. 211–226. Springer, Heidelberg (2000) 46. Peine, A.: Technological paradigms and complex technical systems - The case of SmartHomes. Research Policy 37, 508–529 (2008) 47. Peine, A.: Understanding the dynamics of technological configurations: A conceptual framework and the case of Smart Homes. Technological Forecasting and Social Change 76, 396–409 (2009) 48. Maglio, P.P., Matlock, T., Campbell, C.S., Zhai, S.M., Smith, B.A.: Gaze and Speech in Attentive User Interfaces. In: Tan, T., Shi, Y., Gao, W. (eds.) ICMI 2000. LNCS, vol. 1948, pp. 1–7. Springer, Heidelberg (2000) 49. Hagel III, J.: Spider versus spider. Mckinsey Quarterly (1), 4–19 (1996) 50. Cusumano, M.A.: Staying power - six enduring principles for managing strategy andinnovation in an uncertain world. Oxford University Press (2010) 51. Sandström, G.: Smart homes and User Values - Long-term evaluation of IT-services. Residential and Single Family Dwellings PhD Thesis. Introduction. Royal Institute of Technology, Stockholm (2009)

Developing Sustainable e-Business Models for Intelligent Transportation Systems (ITS) Konstantinos N. Giannoutakis and Feng Li The Digital Economy Hub & The Business School, Newcastle University, Newcastle upon Tyne NE1 7RU, UK {kostas.giannoutakis,feng.li}@ncl.ac.uk

Abstract. Intelligent Transportation Systems (ITS) are technologies for efficient traffic management, e.g. accident and congestion prevention, reduced emissions, traveler information etc. Despite the potential benefits identified by previous research, there is a remarkable lack of a robust e-Business case for ITS. Although the necessary technology exists and the internet provides opportunities for more cost effective integration of systems and services, many ITS are not sustainable and great ideas do not develop further. Many applications are still too expensive to purchase and install, they require substantial upfront investments and the returns will take many years to materialize. This paper emphasizes the need for ITS to take advantage of the developments in Information and Communication Technologies (ICTs) and the internet, and proposes some e-Business model framework for ITS. We also address sustainability issues and aspects of the network economy for ITS companies. Finally, we present a stakeholders’ analysis and a future research agenda. Keywords: Intelligent Transportation Systems, e-Business model framework, sustainability.

1

Introduction

ITS were invented to improve transportation systems operations, by increasing productivity, improving safety, reducing travel time and costs and saving energy [1]. Such systems also open up significant opportunities for new services for drivers, travelers, as well as infrastructure providers. ITS is an umbrella term that cannot be limited to a certain number of systems and applications. There is no commonly accepted taxonomy of ITS, as there are hundreds of systems and applications, each of them designed for a very specific purpose, as well as, a growing number of new devices, or, devices under development. However, attempting to categorize it, we could distinguish ITS into two wide groups: intelligent infrastructure and intelligent vehicles. This taxonomy was introduced in the early 1990s, when the U.S. Department for Transportation adopted the “Intelligent Transportation Systems” project [2]. Intelligent infrastructure includes systems and applications designed for road and transportation infrastructure, T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 200–211, 2011. © IFIP International Federation for Information Processing 2011

Developing Sustainable e-Business Models for Intelligent Transportation Systems (ITS)

201

such as loop detectors, electronic toll collection (ETC) and variable message signs (VMS). The term “intelligent vehicles”, on the other hand, refers to in-vehicle systems and applications, such as satellite navigation devices, intelligent speed adaptation (ISA), adaptive cruise control (ACC), forward collision warning (FCW), pedestrian detection systems (PDS) and lane departure warning (LDW). Despite the fact that a few studies have discussed issues of sustainability in the broader transportation sector [3, 4], not enough research exists on ITS sustainability. A systematic literature review on ITS showed us that the business side of ITS has been ignored, as the vast majority of journal papers target ITS from an engineering perspective. Search of journal databases, such as the Web of Knowledge and Business Source Premier, over all time periods, provided us with a very small number of business related studies on ITS, and no recent papers on business, or e-Business models. The systematic review was based on the algorithmic methodology proposed by Lee [5], which produces a large number of potential references by searching journal databases using specific keywords, and then, eliminates less relevant publications during filtering stages. The main aim of this paper is to provide a robust e-Business case for the sustainable development of ITS. This research is part of a multi-million research project funded by the Research councils of the United Kingdom (RCUK), aiming at improving social inclusion with the use of digital technologies in many facets of everyday life, including transportation. The paper is organized as follows: in the next section we present a brief overview of ITS. Section 2 presents past business studies on ITS and transportation and shows the apparent lack of business model understanding in the literature. Our analysis on a sustainable e-Business model for ITS is given in section 3. Section 4 gives a preliminary stakeholders’ analysis, section 5 highlights some ideas on future research, and finally, section 6 presents our conclusions.

2

The Lack of a Sustainable Business Model

In the last 40 years ITS has been mainly studied from the technological perspectives, and despite previous research and experiments, it has so far failed to take off. The technological possibilities have failed to be translated into real benefits to providers, users and to society. With rapid technological developments, in ICTs in particular, the time has come for ITS to take off. One of the main barriers is the lack of a robust business case, shown by the lack of previous studies in business and economics. The only recent work on e-Business models in ITS and transportation we identified is the paper by Zografos et al. [6], which discusses business models for flexible passenger transport services (FPTS), i.e. customized transportation services characterized by special features, such as options on type of vehicle to use and online booking selection of routes. Zografos et al. [6] present a methodology for ranking and selecting the most appropriate business model, based on criteria such as, legal and regulatory framework, market opportunities, business vision and business mission. Most of the

202

K.N. Giannoutakis and F. Li

business related studies on transportation and ITS, however, do not even discuss business models. They are focused on describing examples of ITS projects and their economic evaluations. Cost-Benefit Analysis (CBA) is the most popular methodology for these economic evaluations. Odeck and Welde [7] use CBA for the automatic Oslo toll collection system, Naniopoulos et al. [8] use CBA to evaluate the results of the TRAVEL-GUIDE project, a European project aiming to provide guidance in the implementation of ITS, Melkert and Van Wee [9] use CBA for the “Superbus”, a conceptual high-speed vehicle for public transportation running from Amsterdam to Groningen. Leviakangas and Lahesmaa [10], on the other hand, highlight the deficiencies of CBA in evaluating and selecting ITS investments and promote multicriteria analysis (MCA), which take into account more factors than CBA, such as experts’ preferences and goal settings. The lack of a robust business case for ITS is evident in most of these studies. The next section describes the e-Business model framework we propose for ITS.

3

Towards an ITS e-Business Model Framework

A business model defines how a business works and the logic that creates its value. It provides the vital links between an organization’s vision and strategy with its structures and processes [11]. The term e-Business model, on the other hand, was first defined by IBM in the 1990s as “the transformation of key business processes through the use of internet technologies” [12]. It is broader than e-Commerce, i.e. buying and selling items over the internet, as it involves customer service, business collaboration and utilization of online services for a wide array of business transactions [12]. Business models are essential for translating commercial opportunities into revenue generating activities, but despite the apparent focus on commercial sustainability, any sustainable business models in ITS need to take into account the divergent nature and unique characteristics of the sector. In particular, the business models need to effectively address the tensions between creating commercial value and generating wide social and economic values; and the distribution of the benefits amongst different stakeholders. There is no commonly accepted definition for a business or an e-Business model. Several authors have studied e-Business models and provided taxonomies based on general features of these models [12]. Although it is difficult to define what an eBusiness model should and should not include, and no universal business model exists one of the most complete e-Business model frameworks has been conceived by Osterwalder and Pigneur [13]. They propose an e-Business model that contains four main pillars: a) Product and services, b) infrastructure and network of partners, c) relationship capital and d) financial aspects. In addition, the sustainability of any business model also depends critically on e) stakeholder credibility, which include both internal and external stakeholders [11]. For any ITS project to be sustainable these elements must be adequately addressed.

Developing Sustainable e-Business Models for Intelligent Transportation Systems (ITS)

3.1

203

Products and Services

Products and services provide the basis for revenue generation and the effects are transversal to several layers within the society. Fig. 1 is presenting the value proposition for ITS. The value creation starts at the level of providing intelligent vehicles and infrastructures, by ITS providers. The created value passes on to the users level in the form of safety, reduced travel time, reduced congestions and all the benefits associated with the use of ITS. This layer includes the direct benefits to the people, for which they are prepared to pay. Hence, that creates business opportunities for more companies to enter the market as mediators and service providers. Next, the value passes to a higher level, namely the “Economy & Society” layer. This is translated into external benefits (e.g. positive economic externalities) such as, reduced spending for new roads, improved traffic management, less congestion, less costs and need for accident response and treatment. It also results in reduced pollution, better social inclusion for disadvantages transport users (e.g. elderly and disabled) and better quality of life.

Fig. 1. Value proposition for ITS

204

3.2

K.N. Giannoutakis and F. Li

Infrastructure and Network of Partners

This element refers to the intelligent infrastructure and the key stakeholders behind ITS development and management. An ITS network could include stakeholders like the government, funding bodies, transport groups, automobile companies, communication technology companies, the energy sector, the road users and others. More about the stakeholders’ analysis of ITS is described in the next section. It is about how the intelligent infrastructure can be used in order to create value. It also shows who the main partners are, what their responsibilities and incentives are and what kind of relationships and partnerships they may develop. The internet provides opportunities to companies to develop e-Business models that make minimum use of physical infrastructure and handle their relationships and contacts with suppliers and partners. ITS companies can benefit from such opportunities, especially in the case of companies that manipulate the digital services of road infrastructure (e.g. variable sign messages, automatic road signs, road cameras, electronic toll collection, satellite navigation systems, and traveler information). However, special attention must be paid to compatibility issues with vehicles. ITS infrastructure incompatible with intelligent vehicles will add no value; it will only mean waste of resources. Therefore, intelligent vehicles should be considered as part of the infrastructure when adopting a particular eBusiness model. That also relies on the following element, the relationship capital, which is about forming a network of customer and managing the relationships with those customers. 3.3

Relationship Capital

This element is about customer relationship and the impact of ITS networks on the road users, how the ITS providers can win the users’ trust on their products and services, how this trust is measured, through which channels the users’ information is collected and utilized, and how the feedbacks from the users are gathered and manipulated by the service providers. Never before had the growth of the internet given the firms such an opportunity to reach a vast network of customers and provide products and services on a global basis at a minimum cost. For ITS companies, the World Wide Web constitutes an ideal marketplace to promote themselves and create rapidly a critical mass for their digitized products and services. 3.4

Financial Aspects

It is about converting the value that the ITS providers offer into revenues and profits, i.e. the pricing models they adopt, how the company makes efficient use of its tangible and intangible assets, how it converts the value of its products and assets into money. In the internet era companies’ interest is shifting more rapidly to investments in intangible assets (e.g. reputation, network of suppliers, intellectual property, value of information), while tangible or physical assets constitute a decreasing percentage of the total company value [12]. This in turn also serves the tasks of cost reduction and of working more effectively with the same or even fewer resources. Some ITS companies are capital intensive due to the nature of their line of business. Although an

Developing Sustainable e-Business Models for Intelligent Transportation Systems (ITS)

205

ITS company might need huge upfront spending and often take a long time to recover the costs and make profits, an e-Business model that takes advantage of the facilitation that the world wide web offers can speed up the process of recovering the costs and making profits. There are numerous examples of online firms that experienced exploding profitability and corporate value within a few years through proficient use of the internet (e.g. Google, Amazon, e-Bay, Facebook etc). The financial aspects sit upon all previous three elements and affect them. It is therefore necessary to secure a sustainable business plan, able to offer satisfactory returns and benefits to the investors and provide assurance that the contributors will be rewarded. Any business plans must also be self-financed after some point. 3.5

Stakeholder Credibility

The sustainable development of ITS also depends critically on the support of all key stakeholders, which goes beyond financial sustainability. A particularly important aspect of the stakeholder credibility in ITS is about the distribution of potential benefits amongst different interest groups, and the wider social and economic benefits at the society level. 3.6

Sustainable e-Business Models for ITS

The main challenge for an ITS company, as implied from the analysis above, is to take advantage of the opportunities offered by the internet. There is growing development in the ITS technology, both for transportation infrastructure and vehicles, and several examples of products and services in the ITS industry that could be integrated with, or make use of the internet and digital technologies. Nevertheless, the ITS sector has not taken off and the opportunities offered by the ICTs have remained rather unexplored. In the following, we present a few examples of ITS technologies, where the internet could create business opportunities. Advanced Traveler Information Systems (ATIS). They are systems that provide customized information to the user, such as on route selection, options about public transport, information about the destination and warning messages for potential dangers during travelling [14, 15]. Many of them are GIS-based, like in-vehicle navigation devices and rely on digital technologies in order to operate. They offer an excellent opportunity for e-Business model development, through for example, offering information on products of third parties (e.g. adverts), or even operating as platforms for added services, such as reporting defects on the road and updating online information about routes. They could also incorporate tourist information for popular destinations, such as suggestions for sightseeing, or, online hotel booking. Electric Vehicles. With the increasing environmental concerns of the future, there is growing research on electric and hybrid vehicles [16]. Digital technologies could contribute to the establishment of online business for these vehicles, for example by providing online information about charging posts, or, by selling online credit for vehicle charging on a pay-as-you-go basis with a registered smartcard.

206

K.N. Giannoutakis and F. Li

Electronic Toll Collection. Electronic Toll Collection (ETC), also known as Electronic Payment and Pricing System, is a topic of growing interest [7, 17, 18]. It is technology that enables the collection of congestion charging automatically, by recognizing the vehicle’s registration number. Neither have the drivers to stop at toll plazas, nor are cashiers required for the collection of tolls. Through the internet, public authorities, or, private companies could enable prepayment of tolls, or even discounts, for frequently used routes, by allowing the road users to set up online accounts through their websites. ETC could also be used as supportive mechanism for tracking down vehicles linked to illegal activities and facilitate law enforcement. Public Transportation. It is one of the areas where ITS have already started to have an impact and to revolutionize public transportation services [19, 20]. Widely used examples of ITS are passenger information systems at bus stops or train stations, busmounted cameras, online bookings and automatic payment systems. There is room for further developments of internet-based technologies on public transportation, such as improvements in the integration of traveler information with mobile technology and enforcement of Wi-Fi networks, which could create new business opportunities for third parties.

4

Stakeholders’ Analysis

In order to map the network of ITS stakeholders in a simplified way, we list the following main players: • • • • • • • • • •

Government & Policymakers (G&P) Funding Bodies (FB) Transport Groups & Organizations (TG&O) ITS Designers & Manufacturers (ITS-D&M) Automobile Suppliers (AS) Key Shareholders (KS) Energy Sector (ES) Environmentalists (E) Local Authorities (LA) Users (U)

We classify government & policymakers, funding bodies, ITS designers & manufacturers, key shareholders and the energy sector as the primary stakeholders and all the rest as the secondary stakeholders. In Fig. 2, we present the influence-interest grid mapping technique, developed by Imperial College London. The horizontal axis measures the level of interest of the stakeholder in the technology. The vertical axis measures how much the stakeholder can influence these developments. Each circle, or bubble, represents a stakeholders group, with red for the primary stakeholders and green for the secondary. The size of the bubble is indicative of the size of the group.

Developing Sustainable e-Business Models for Intelligent Transportation Systems (ITS)

207

Fig. 2. Influence-interest grid mapping for ITS stakeholders

4.1

Government and Policymakers (G&P)

A rather small not-for-profit group, with moderate interest in ITS limited to the extent that ITS could be a source of public funds or a means to more efficient traffic management. They can have great influence for the implementation of ITS projects as they have legislative power under their control. However, their influencing power might be limited, because of conflicting interests with other key stakeholders, such as funding bodies and the energy sector. 4.2

Funding Bodies (FB)

It is also a small group, but with great influence on ITS implementation, as they are the ones who provide the funding. They could also be characterized by increased interest as long as they foresee significant returns in their investments. They have overlapping areas and collaborations with the Government & Policymakers and the key Shareholders. 4.3

Key Shareholders (KS)

They have many common characteristics with the previous two groups in terms of interest, influence and size. They also constitute another source of funding and that assigns them both with high influence and interest.

208

4.4

K.N. Giannoutakis and F. Li

Transport Groups & Organizations (TG&O)

A middle size group with moderate interest in ITS, as long as ITS does not threat vested interests. They have not much level of influence either, as their role is usually restricted in researching the transportation field or creating revenue by entrepreneurship in transport operations (e.g. university research centers, public transport cooperatives, freight transport or leisure transportation companies). 4.5

Automobile Suppliers (AS)

They are very interested in the ITS developments, as they can change radically the market conditions they operate in, but have no much influence in these developments. Their role is mainly to intermediate between the manufacturer and the final customer. They have some overlapping areas and collaborations with designers and manufacturers. 4.6

ITS Designers and Manufacturers (ITS-D&M)

They are amongst the most interested groups in ITS developments, in many cases they are the driving force of these developments, but with very little influence in getting ITS out in the market. 4.7

Energy Sector (ES)

They are not interested in a positive way in ITS, especially if they compromise their interests in the energy supply (e.g. electric vehicles that render petrol obsolete). Nevertheless, due to their financial power, they might exert large influence to the ITS developments, either positively, or negatively. 4.8

Environmentalists (E)

A rather small group, with moderate interest in ITS and very little influence as well. Their ITS interest is sparkled either if ITS poses threats to the environment, or, if it is deployed for more environment-friendly transport. In any case, their power is very limited. 4.9

Local Authorities (LA)

A large group with also moderate interest in ITS and average power to influence the developments. In many countries their power is restricted to executing the central government’s decisions. 4.10

Users (U)

They are the largest group of stakeholders. They sit somewhere in the middle of the grid, meaning that in most of the cases they have a passive role in ITS developments.

Developing Sustainable e-Business Models for Intelligent Transportation Systems (ITS)

209

They are interested in ITS according to their needs for transport, but they do not have much influence. They are affected by all stakeholder groups, but they have closer interactions with the Transport Groups & Organizations and the Local Authorities. The mapping we presented is based on the evidence we found in the literature. Part of our empirical research is going to provide a real mapping of such stakeholders, which to the best of our knowledge, has not been attempted yet. Such a piece of research could be extremely valuable in our efforts to establish a holistic framework for the ITS. It will also provide a real picture of the industry and will initiate comparisons between the current state-of-the-art and the ones that should be targeted.

5

A Future Research Agenda

Researching ITS as a whole has provided us with useful insight. It also highlighted the absence of a sustainable business model framework, regarding a wide spectrum of systems and applications with increasing popularity. Nevertheless, as mentioned in section 2, ITS is a very broad term, each category of ITS applications has its own special characteristics and involves the action of specific groups of stakeholders. The e-Business model framework presented in this paper can serve as a generic guideline for constructing future business models, but empirical research is also necessary in order to facilitate the full-scale development of the field. There are various ITS examples that provide fertile ground for e-Business research, e.g. advanced traveler information systems (ATIS), electric vehicles, wireless networks for inter-vehicle communication, public transportation, and we have shown examples how some of them can operate as platforms for third parties’ services. Empirical research will need to narrow down to particular examples and collect data with several types of research. Three types of research are needed: surveys to understand general patterns and emerging trends; case studies to develop deep insights; and simulation to build on surveys and case studies and allows us to explore different scenarios and generate insights in different contexts. • Surveys: Both online and by contacting real-life key stakeholders, like system designers, traffic authorities, users, policymakers, funding bodies, academics etc. These contacts could be through interviews or by using questionnaires. Through surveys, useful information can be extracted on the market size, current business strategies and on the key stakeholders in the market. • Case studies: Targeting of specific companies who have excelled in the internet economy and careful examination of their e-Business models. This part of research could focus on companies that operate in the transportation sector, but it could be expanded to companies of other sectors as well, whose e-Business models can be adapted to the ITS industry and expanded. • Simulations: Another important part of our research would be simulated studies of particular scenarios. Simulated firms and stakeholders will provide realistic information on business behavior under certain circumstances.

210

6

K.N. Giannoutakis and F. Li

Conclusions

This paper addressed business and economic issues of ITS and discussed aspects of developing a sustainable e-Business model framework. We also presented a preliminary stakeholder analysis, on which we would like to develop our empirical research. The business model framework we proposed consists of five main elements: products and services, infrastructure management, customer relations, financial aspects and stakeholder credibility. Empirical research is necessary in order to gather primary data from businesses and social groups, able to inform any future research avenues through surveys, case studies and simulations. Acknowledgements. We acknowledge the financial support by the Social Inclusion through the Digital Economy (SiDE) project at the Newcastle Digital Economy Hub, funded by Research Councils UK (RCUK) under its Digital Economy Research Programme.

References 1. Miles, J., Chen, K. (eds.): The Intelligent Transport Systems Handbook, 2nd edn. The World Road Association (PIARC), Swanley (2004) 2. Wootton, J.R., Garcia-Ortiz, A., Amin, S.M.: Intelligent Transportation Systems - A Global Perspective. MComM 22(4-7), 259–268 (1995) 3. van den Bergh, J., van Leeuwen, E.S., Oosterhuis, F.H., Rietveld, P., Verhoef, E.T.: Social Learning by Doing in Sustainable Transport Innovations: Ex-Post Analysis of Common Factors Behind Successes and Failures. Res. Pol. 36(2), 247–259 (2007) 4. Cohen, M.J.: Destination Unknown: Pursuing Sustainable Mobility in the Face of Rival Societal Aspirations. Res. Pol. 39(4), 459–470 (2010) 5. Lee, R.: Social Capital and Business and Management: Setting a Research Agenda. International Journal of Management Reviews 11(3), 247–273 (2009) 6. Zografos, K.G., Androutsopoulos, K.N., Sihvola, T.: A Methodological Approach for Developing and Assessing Business Models for Flexible Transport Systems. Transportation 35(6), 777–795 (2008) 7. Odeck, J., Welde, M.: Economic Evaluation of Intelligent Transportation Systems Strategies: The Case of the Oslo Toll Cordon. IET Intelligent Transport Systems 4(3), 221–228 (2010) 8. Naniopoulos, A., Bekiaris, E., Panou, M.: Cost and Benefits of Information Technology Systems and Their Application in the Infomobility Services: The Travel-Guide Approach. Economic Impacts of Intelligent Transport Systems: Innovations and Case Studies 8, 463– 480 (2004) 9. Melkert, J., Van Wee, B.: Assessment of Innovative Transport Concepts Using CostBenefit Analysis. Transportation Planning and Technology 32(6), 545–571 (2009) 10. Leviakangas, P., Lahesmaa, J.: Profitability Evaluation of Intelligent Transport System Investments. Journal of Transportation Engineering-ASCE 128(3), 276–286 (2002) 11. Li, F.: What Is E-Business and Does It Still Matter? International Journal of Information Technology and Management 6(2-4), 125–137 (2007) 12. Li, F.: What Is E-Business? How the Internet Transforms Organizations. Blackwell Publishing, Oxford (2007)

Developing Sustainable e-Business Models for Intelligent Transportation Systems (ITS)

211

13. Osterwalder, A., Pigneur, Y.: An E-Business Model Ontology for Modelling E-Business. In: Bled 2002 Proceedings, Paper 2, Bled, Slovenia (2002) 14. Singh, V., Kumar, P.: Web-Based Advanced Traveler Information System for Developing Countries. Journal of Transportation Engineering-ASCE 136(9), 836–845 (2010) 15. Yang, H., Huang, H.-J.: Modeling User Adoption of Advanced Traveler Information Systems: A Control Theoretic Approach for Optimal Endogenous Growth. Transportation Research Part C: Emerging Technologies 12(3-4), 193–207 (2004) 16. Yang, C.J.: Launching Strategy for Electric Vehicles: Lessons from China and Taiwan. Technol. Forecast. Soc. Change 77(5), 831–834 (2010) 17. Chen, C.D., Fan, Y.W., Farn, C.K.: Predicting Electronic Toll Collection Service Adoption: An Integration of the Technology Acceptance Model and the Theory of Planned Behavior. Transportation Research Part C-Emerging Technologies 15, 300–311 (2007) 18. Lee, W.H., Tseng, S.S., Wang, C.H.: Design and Implementation of Electronic Toll Collection System Based on Vehicle Positioning System Techniques. Computer Communications 31(12), 2925–2933 (2008) 19. Velastin, S.A., Boghossian, B.A., Lo, B.P.L., Sun, J., Vicencio-Silva, M.A.: Prismatica: Toward Ambient Intelligence in Public Transport Environments. IEEE Transactions on Systems Man and Cybernetics Part A - Systems and Humans 35(1), 164–182 (2005) 20. Vanajakshi, L., Subramanian, S.C., Sivanandan, R.: Travel Time Prediction under Heterogeneous Traffic Conditions Using Global Positioning System Data from Buses. IET Intelligent Transport Systems 3(1), 1–9 (2009)

Observations on e-Business Implementation Capabilities in Heterogeneous Business Networks Tero Pesonen and Kari Smolander Lappeenranta University of Technology, Skinnarilankatu 34, 53850 Lappeenranta, Finland {Tero.Pesonen,Kari.Smolander}@lut.fi

Abstract. The objective of this qualitative study was to understand the factors that hinder large industry organizations in implementing e-Business solutions. The study included 4 organizations that operated in heterogeneous business networks consisting also of small partners. Interviews were conducted with managers of IT, purchasing and business development. The results indicate that the organizations faced a mixture of intertwined, accumulative challenges. Lack of IT resources and growing reliance on third parties had reduced IT expertise and made investments difficult. At the same time, e-Business projects were found to be very demanding to plan and execute, and with no e-Business knowledge or conscious e-Business strategies in place, the organizations were progressing haphazardly. The heterogeneous business environment also posed external challenges with partners having highly-disparate IT capabilities. Existing e-Business technologies appeared either inadequate for the needs of these networks or the appropriate technology was entirely missing. Keywords: e-Business, e-Business implementation, interoperability, IT implementation, inter-organizational systems.

1

Introduction

E-Business can be defined as the use of information technology in support of more efficient business interactions between two or more organizations or organization units [6]. e-Business therefore relates not only to specific information technologies, but also to collaboration, either internal or inter-organizational. According to traditional collaboration schemes, transaction-oriented relationships between organizations in supply chains have led companies into trying to take advantage of one another in order to maximize their own utility. In such a context, companies see themselves as buyers of goods and sellers of value-added goods [2]. The first eBusiness applications, based on the EDI (Electronic Data Interchange) standard [3, 4, 5], aimed to make these basic, high-volume interactions taking place between business partners – for example, a buyer and seller – electronic and more automated. EDI standards replaced print documents sent by mail with electronic documents exchanged directly between IT systems. This allowed for removing manual work and errors and for reducing clerical costs [6, 7, 12]. Modern e-Business technologies, such T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 212–226, 2011. © IFIP International Federation for Information Processing 2011

Observations on e-Business Implementation Capabilities

213

as RosettaNet-like business-to-business integration (B2Bi) frameworks, have since built on these objectives by offering richer, platform neutral business data formats and support for fully-automatic business processes that can be integrated between collaborating partners [1, 8, 9]. Recently, Web Services and other cloud computing technologies have also emerged, but not so much as direct e-Business tools or applications as alternative e-Business development platforms that allow software to span traditional organizational boundaries [10, 11, 34]. Despite these technological developments, e-Business has failed to reach the level of sophistication and adoption originally predicted [6]. While the use of internet for business has become ubiquitous in developed economies, and wholly new business models, such those made prevalent by Google and other internet-based companies, have become pervasive [24], the actual business interactions between organizations have changed slowly [13]. Deep supply chain collaboration has failed to materialise [30], and still only a portion of the industry makes use of any kind of e-Business technology at all [6, 13, 14]. In this study, our objective is to observe large industry organizations in order to understand what the factors are that hinder them in implementing e-Business solutions for more efficient B2B integration. The organizations represent traditional, non-IT industries and operate in heterogeneous business networks that comprise also smaller suppliers and buyers. Our study is qualitative, as understanding e-Business implementation constraints requires exploration in socio-technical real-world contexts. We have structured our paper as follows: We first review the related research in chapter 2. We define our research objective and explain our research process in chapter 3. We then present our findings in chapter 4 and conclude our paper by discussing and summarizing the results in chapter 5.

2

Related Research

One of the drivers of modern e-Business research is the inefficiency caused by lack of interoperability, shown first as countable costs by [31, 38] in a series of studies concerning the American auto industry. According to Legner & Lebreton [39], interoperability research can be roughly classified as either explorative, which involves identifying interoperability factors or describing the status of interoperability in a given domain, or constructive, which involves developing solutions for overcoming interoperability challenges. Modern e-Business interoperability research often focuses on SME's (Small and mid-sized enterprises), whose slow e-Business adoption [13, 14, 17] has received a lot of attention. This is because SME's comprise the majority of enterprises in most economies and hence have a major influence on the economy’s efficiency (OECD [49]). On the other hand, studies in e.g. the UK, Sweden, and New Zealand, [15, 16, 21] respectively, have shown that SME's are not a homogeneous group but vary greatly both in e-Business use as well as in the factors that influence them in adopting or rejecting e-Business technology. Moreover, these factors do not necessarily reflect

214

T. Pesonen and K. Smolander

the companies' business strategy or objectives any way, but may relate to e.g. an individual manager's influence [15], or to available resources and know-how [32, 33, 35]. Studies in developing economies (e.g. [34, 35]) agree for most part with these findings and emphasize that each SME should be viewed as unique with its own set of e-Business implementation issues [6]. Research on e-Business adoption at large and multinational companies (MNC's) is more limited and tends to also be more industry-sector specific. Machner et al. [45], Hetemäki & Nilsson [47], and Mustonen et al. [48], for example, have analyzed eBusiness drivers specifically in the semiconductor and forest industry. Although industry-specific, the findings reflect MNC's as more interested in a strategic, long term business-integration-driven approach that goes beyond single e-Business applications. Direct comparative studies between SME's and large enterprises (e.g. [46, 50]) consider much lighter e-Business approaches, such as various kinds of web commerce. The studies indicate that SME's are more nimble and flexible adopters of light-weight e-Business applications, and seem to benefit from investing in them, sometimes even more than their larger counterparts. The use and development of the more demanding B2Bi technologies is, however, typically restricted to larger organizations, which have more resources and more opportunities to profit from supply chain integration [1, 6, 22]. This is further confirmed by earlier EDI studies [36, 37] and claimed by the RosettaNet consortium [40]. Constructive e-Business research remains also SME-oriented. Here the European Union funded research projects are most notable. The ATHENA project [18] (built on a number earlier projects) was a major effort to build a common European enterprise interoperability framework that heeds not only ICT, but also knowledge, business and semantic interoperability, and could therefore be used in gathering further research results and for developing interoperable services. It remains unclear, however, what concrete outcomes the ATHENA platform has contributed. The ITAIDE [41] project has proposed cross-border e-Business (eGovernment) solutions while the GENESIS project [42] developed an architecture and a process-and-document modeling approach for allowing SME's to carry out e-Business activities with other SME's, public sector and financial institutions via the GENESIS system. It appears that no SME-to-MNC B2Bi is directly supported. Finally, efforts to simplify e-Business standards testing have been also carried out in global co-operation [20]. According to Legner & Lebreton [39], analysis on the strategic, organizational and operational issues that relate to interoperability are still missing despite the gamut of research already carried out. The authors argue that although projects such as ATHENA have established a foundation for interoperability research, the focus still remains too technology oriented. By the same token, Smolander & Rossi [43], who observed an enterprise-wide e-Business architecture development process, found the interplay between the architecture and the organization, and the inevitably raising conflicts and compromises, as more important to manage than mere technical factors. The distinction between enterprise architecture [44] and modern, large e-Business architecture appears increasingly intangible if not already non-existent. Nevertheless, in view of the more recent COIN and COMMIUS projects [19, 26], the technologydriven approach persists at least in current interoperability research.

Observations on e-Business Implementation Capabilities

3

215

Research Process

Our initial objective was to study e-Business implementation requirements for specific supply chain networks. We quickly found, however, that the body of knowledge in this area was limited. We were interested in large organizations that operate in heterogeneous business networks comprising also a number of much smaller suppliers and buyers with highly-disparate business profiles, IT investment capabilities and IT implementation skills. As we discuss in our review of the pertinent literature, e-Business research has mainly focused on SME-to-SME or otherwise very homogeneous, “flat”, domains, treating small and large organizations often separately or disregarding the concrete constraints that B2Bi activities may impose on a heterogeneous business environment. As an example of this, a large paper mill attempts to leverage e-Business solutions, but does not succeed because the majority of its partners, who are SME's, are unable to implement and operate the technology. Our research objective is, therefore, to understand what are the factors that hinder large traditional industry organizations in implementing e-Business solutions. More precisely, we want to investigate and characterize the issues present in these organizations or their business environments that have a negative effect on the organizations' ability to develop or adopt new B2Bi solutions or to make effective use of existing ones. In particular, we want to understand the role of the organizations' business environment in e-Business adoption – our study observes a heterogeneous business network consisting also of small partners. This was in our view necessary in order to understand all the dimensions of real world B2Bi schemes. 3.1

Research Methodology

e-Business implementation is a process of developing complex, socio-technical systems that involve various stakeholders, IT resources, business processes and objectives influencing one another. Understanding this process requires exploration in real world contexts, which suit well for studying with qualitative methods [27]. We used as our qualitative research approach Grounded Theory (GT) as described by Strauss and Corbin in [23]. Originally described by Glaser and Strauss in [28] for social sciences, the method has since been employed also for research in various fields owing to its ability to discover novel concepts and develop new theories. GT bases the theory building on induction: Instead of encouraging apriori theorization and hypotheses, the method rather instructs the researcher to start directly with data collection and to arrive, by means of explicitly-specified analytic steps, at a specific theory only based on what has been developed from the data, not on what has been expected or wanted to prove with the data. The first analysis phase, open coding, involves discovering and labeling concepts that might denote commonalities in the data; concepts that stand for “what is going on” in the part under analysis are considered particularly fruitful for further scrutiny. The concepts are then assorted by assigning them properties. In the second phase, axial coding, the concepts are further developed by defining their relationships. A crucial tool for this is various comparisons that aid in discovering commonalities and differences, and the properties of these variations. GT provides a set of mechanisms

216

T. Pesonen and K. Smolander

for carrying out these comparisons, but it is always the researcher's creativity that in the end plays the main role in the analysis. The third and final phase is selective coding, where a final core category is derived from the fully developed categories and thus theory is created. 3.2

Case Organizations and Interviewees

Our first criterion in selecting the case organizations was that we wanted them to be located in a limited geographical region where the number and type of potential suppliers as well as the operational environment were similar. The organizations also had to be large enough to be able to impose a notable influence in the region by attracting around them a network of smaller suppliers specialized in serving large industry. Since the heterogeneity of business partners was an important factor for us, we decided to include one large public sector organization along with industry representatives. Existing research has paid some attention to the needs of public sector in relation to e-Business development (e.g. [22, 29, 42]), but not specifically in the context of carrying out day-to-day business interactions – for example, as a purchaser of services and supplies. On this score, we treated the public sector organization similarly to the industry representatives. Table 1 lists the case study organizations and the position of the interviewees in each organization. Nine manager-level interviewees were selected from both business and IT departments based on preliminary discussions with the organizations. The interviewees were carried according to a list of prepared, open-ended questions that were similar to all participants albeit with slight differences in emphasize depending on the interviewee's background. In total, we accumulated approximately 11½ hours of interview material, which was then transcribed to text for analysis. In addition to the interviews we also participated in a meeting involving the managers in Organization D interested in hearing and discussing e-Business issues in relation to their newly-started ERP development project. We used this opportunity for questioning and for observing the discussion. Table 1. Case organizations Label A

B C D

Description City organization. Provides public services.

Interviewee IT manager Head of Purchasing Financial director A paper and pulp mill part of a global Head of maintenance services wood processing company. Head of mill services A wood products mill part of a global IT manager wood processing company. Head of purchases A steel mill part of an international Head of business developments consortium. IT manager Meeting with other managers

Observations on e-Business Implementation Capabilities

4

217

Findings

Figure 1 depicts our observations of how specific impeding factors affect the organizations' e-Business implementation capability. We have organized the specific factors under three generic classes – IT resources, e-Business knowledge and Business environment, of which IT resources and e-Business knowledge are internal to the organization and in its own control. Each high level category manifests the implications its sub factors impose. The high level factors are ordered according to importance from left to right so that IT resources has, in general, more severe impact than Business environment. The influence between categories and individual factors is marked with arrows.

eBusiness implementation obstacles Internal

External

IT Resources

eBusiness knowledge

Business environment

Inability to take initiatives

eB projects are difficult to manage

Current eB technolo gies fit poorly

Outsourcing

Haphazard eB development steps

Resistance to change

Limited IT expertise

Lack of eBusiness knowledge

Insufficient partner knowledge

Insufficient IT resources

Disparate eB Imple mentation capabilities

Fig. 1. Factors impeding e-Business implementation

The model was constructed to present the highest level categories resulting from selective coding so that IT resources, e-Business knowledge and Business environment remained the topmost subcategories of the core category e-Business implementation obstacles. Similarly, each subcategory contained a set of its own sub categories, labeled here as factors. The analysis started with no explicit a priori categories. The interview data were first scanned for words and phrases describing interesting phenomena or common occurrences, with particular emphasis on activity – “what is happening here?” or “what action does this pertain to?”. With a number of such labels discovered, we were able to categorize them with comparisons, creating new categories, such as “manual procurement process” and “electronic procurement process”, which, while belonging

218

T. Pesonen and K. Smolander

to a yet higher level category, differed on a key dimension of one of their shared properties. We carried both open and axial coding more or less simultaneously. In developing the core category and our theory, we abstracted the categories by developing their relationships until we had arrived at 11 factors, belonging to the three high level categories. We derived the factor relationships, shown in the model by arrows, from the category relationships by asking “Do the properties of this category beget any of the major properties of another category?” We explain our findings according to this model. First, however, we describe the organizations' business environment and e-Business implementation status at the time of the study in order to set the context for our model. 4.1

The Current Status of e-Business in the Case Organizations

The business environment is highly heterogeneous. All case organizations operate not only with other large buyers and suppliers, but also with a network of local SME's that are specialized in offering the kind of supplies and services the local large industry needs but does not produce itself. These supplies can be divided into materials and products (Organization A) and into maintenance work and related services (Organizations B, C and D). Other kind of out-sourced services, such as logistics or IT, are primarily acquired from larger providers outside the region. The business environment is increasingly competitive. The industry organizations (mills) export almost all their production. Competition in their export markets has increased noticeably with globalization, forcing the companies to change their business strategy. B and C have launched drastic cost cutting programs to regain operational efficiency. Organization D, in contrast, has steered away from direct price competition and chosen as its strategy to make special-grade, per-client-tailored products that also incorporate services for additional value. All organizations have chosen to outsource more and more of the supporting day-to-day business operations (e.g. maintenance work, IT, logistics). The organizations have successfully adopted electronic invoicing. Electronic invoicing solutions replace printed invoice documents, traditionally sent via mail, with electronic documents that can be exchanged between information systems and processed digitally. For the case study organizations, electronic invoicing has become pervasive quickly. During the last two years organizations B, C and D in particular have grown the share of eInvoices to more than half of all invoices sent or received; and B and C will soon start requiring all their purchases to be invoiced electronically. A and D, in contrast, do not yet want to impose such demands but are rather content to only promoting eInvoicing to partners and offering help in its adoption. The objectives in using eInvoicing are improved efficiency via reduced manual work and better transparency via more integrated, automated business processes. It also allows for integrating the handling of invoicing data with enterprise resource planning (ERP) systems, which makes it possible to monitor work flows more easily and thus make orders and their status more transparent and easier to predict and plan.

Observations on e-Business Implementation Capabilities

219

By using their influence, the large organizations have not only been able to persuade, but in the case of B and C and partly D, have also been able to dictate the rules by which this technology is to be used in the immediate business network for their benefit. Use of other forms of e-Business was limited. Organization C has made some use of the forest-industry-specific B2Bi standard in business document exchanges involving other large industry partners. The company has also worked with large partners on tailored interfaces that offer access to its IT services. Organizations A and B have attempted to develop e-Business portals for interactions with SME partners, but these projects have either been postponed due to budget issues or they have been incorrectly focused due to insufficient requirements analysis. 4.2

Impeding Factors for e-Business Implementation Capability

IT Resources Insufficient IT resources are a major e-Business implementation obstacle. The costcutting strategy the industry organizations in particular have entered has applied also to IT. This has led every organization to outsourcing all but the most essential day-today IT administration and development activities. The trend has become pervasive: Even for IT tasks not (or not yet) outsourced, the staff has been or is to be reduced as low as possible and may be moved off-site for consortium-wide, centralized IT management – this is particularly true in organizations B and C – and where special expertise is to be needed, it will be acquired from outside as necessary instead of keeping (expensive) IT expertise on a regular pay-roll. This has limited the IT expertise the organizations have at their disposal. “Within this organizational unit, there is only I and [another manager name] left who can any more give answers to such questions [pertaining to IT and e-Business].” “We've been talking about this [outsourcing and centered IT management] for some years already... the objective of course is to make everything a lot easier, so we need to keep fewer IT experts. And we need not hire new IT people when employees retire. Yes, cutting costs is driving all this.” “Instead of doing these [IT tasks] ourselves, we can now buy them as a service from [provider].” Outsourcing. While believed to cut costs without impeding on IT capability, outsourcing of IT appears to have made, with less in-house IT expertise available, the organizations increasingly dependent on third party providers, who have thus gained more control on what kind of systems are used and how they are built and tailored. These providers do not necessarily share the same interests as defined in the organizations' IT strategies, and do not hence promote new IT paradigms, such as eBusiness technologies. Rather, they may support the development of systems that make e-Business-related integration tasks actually more difficult due to undocumented interfaces and missing planning for future expandability. There was evidence of this already having an effect.

220

T. Pesonen and K. Smolander

“We are sure [the system provider] will take care of these aspects... that the XML support will be OK in the new system. I am sure they understand this need and know how to handle it [according to our requirements now and in the future].” These needs related to the XML-support had not been communicated to the provider, nor was the person able to express them in anything but very vague terms. “The problem with this [provider] system is that it is difficult to make it work for this [e-Business use], as it is has always been difficult to read data from it [in any other way than the provider wants] and even today we find it difficult to write data into the database component.” These problems had affected this organization's earlier e-Business project. Inability to take initiatives. Reduced IT resources and the reliance on outsourcing has led to an inability to invest in long term objectives and to discover and take initiatives whose goals lie beyond the day-to-day “sphere of focus” – that is, to make use of IT proactively in order to create new usage scenarios or to support the business strategy in novel ways. This contributes in particular to issues discussed under “eBusiness knowledge” This is also in contrast to organizations B and C's past strategy for developing IT as a long term tool for achieving the industry leadership. “The primary objective for any IT project we can implement today [heeding the costs-saving strategy] has to be... it has to be able to not only to remove costs, or to make something “better”... that is not enough... the project has to result in jobs being eliminated... now. Unless you can show that, it won't be financed.” e-Business Knowledge Organizations are lacking e-Business knowledge and therefore progress haphazardly with no conscious e-Business strategies defined. Lack of e-Business knowledge. Our interviews showed that the notion of e-Business was not only vague and poorly understood, but it was also treated disparately amongst the interviewees. Although all interviewees were somewhat familiar with the term or its variants, each gave a unique definition for what they believed e-Business meant. For example, e-Invoicing, which all interviewees were familiar with, was treated as a separate technology and often only associated with other e-Business technologies upon questioning on possible links between the organization's conceivable future eBusiness objectives. In general, the interviewees approached the topic from a point of view closely tied to their past, professional experience and their current position in the organization. “In my view e-business is... I understand it as belonging to the marketing side... As in our industry there was this hype some 10 years ago that all buying and selling would go to online auction and market sites.” “It is, I think, transmitting data in an electronic form via internet. So there is less need for manual work when the data are transmitted. And it enables huge data storages which people can peruse.” “It means that we make the transactions between our partners electronic – between both clients and suppliers. This can be done in two ways: [long technical explanation omitted]”

Observations on e-Business Implementation Capabilities

221

Haphazard e-Business development steps. Lack of e-Business knowledge has resulted in no conscious organization-wide e-Business strategies defined. The organizations were rather “groping” their way toward something undefined in small, haphazard steps rather than following a organizationally-agreed-on path leading towards specific objectives. The ability to discuss and introduce e-Business-related topics to day-to-day business planning relied on few, select individuals and their interest or motivation to work on the topic and influence the organization. Since eBusiness had remained so little understood, managers could also easily disregard it with impunity in view of their superiors and peers. Lack of e-Business knowledge and organizational understanding on its implications had together further led to the incorrect view that e-Business is simply an IT topic, while in fact it involves, due to its spanning the whole organizational architecture, inevitably not only IT, but the whole gamut of business processes, related business objectives and information systems. Successful e-Business development has to heed know-how across business units and involve the knowledge of domain experts. Consequently, e-Business projects are difficult to manage. The danger that an eBusiness project fails due to improper requirements analysis seems alarmingly high. Organizations C and B in particular expressed great concern over this matter and even appeared wary of starting e-Business initiatives due to the experience of how difficult it is to manage e-Business requirements. For organization A, an e-Business project being developed at the time of the study appeared to be incorrectly focused due to a flawed and partially-skipped analysis phase. The analysts had incorrectly relied on their previous business and IT experience and this way made innocent-looking, “selfevident” presumptions that were inappropriate in an e-Business domain. “That project they are developing for us... well it is completely useless from our requirements point of view. We would rather need the catalog service kind of solution … Lacking it is actually a bottleneck for us.” “The [light-weight e-Business project the mill has tried to develop] has been again postponed; now it was piloted but the pilot failed – the system had been specified incorrectly. They've tried to fix it again and again but they don't seem to get it working right … and it is now being re-analyzed for the requirements and then we reimplement it, but only in case it won't be too expensive.” Business environment The organizations' business network poses external factors that make implementing eBusiness solutions more difficult. A heterogeneous business network consists of partners that have disparate eBusiness implementation capabilities. This can be a major hindrance. SME partners especially are notable here, as the interviewees had a strong opinion that their SME partners would be incapable of adopting such e-Business technologies as would allow the large organizations to make also other transactions than just invoices electronic. In these transactions product orders, inquiries, confirmations etc. would be exchanged as standardized, electronic business documents between each partners' information systems. Electronic invoicing solutions have been able to overcome this barrier because there were third party services available which the SME partners could tap

222

T. Pesonen and K. Smolander

into in case they were unable or unwilling to implement the technology. There were no such solutions available for other, more demanding B2Bi technologies. Other large organizations can also pose problems. As C, for example, has experienced, there is no guarantee that the other partner, no matter what size, is yet capable or willing to participate in electronic transactions beyond eInvoicing. Individual managers may wield significant power over such decisions. Their influence is fueled by the already mentioned factors of lack of e-Business knowledge and lack of e-Business strategies. Insufficient partner knowledge. The organizations are poorly acquainted with most of their partners' IT capability and e-Business-related needs. This is particularly true with SME's, but it is also a limitation with larger partners. In the past, the organizations have not needed to develop this kind of knowledge. But e-Business solutions require for most profit for the partners to co-operate, and lack of knowledge on possible mutual interests, or areas where both organizations or even a group of organizations could benefit from working together on e-Business implementation, affects the network's ability to collaborate. “About our SME partners... well, we know nothing of what is going on there, what they can do, what they need. It is a completely blank area for us!” There is also resistance to change in the business network. This is, in part, attributed to a kind of chicken and egg situation: There is limited motivation even for a large organization to invest in and develop capability for something that others are not (apparently) yet doing – the fear is it might turn out to an entirely wrong standard or approach as the technology “catches on someday.” Other organizations may also want to pursue only their selected e-Business standard. “We see little activity regarding this amongst our partners... I feel they tend to think, let's wait and see. So there is perhaps some resistance to change due to our industry sector.” “It may be the partner already has another standard in use, and if they are a customer... a large one... we just have to go by that standard and implement a mapping between that and our own.” “For some customers our industry is not their primary business, so they see no incentive to implement our industry standards.” The mills thought that in their traditional industry sectors it would be difficult to gain competitive advantage with e-Business technologies today. They also pointed out that their industries are rather followers than early adopters or IT technology drivers. Therefore competitors – internal or external to the network – were not expected to make any quick, major breakthroughs either, or gain unexpected competitive advantage through B2Bi. “This [e-Business development] is 'business as usual' for us [in our industry]. No one here gains a competitive advantage from it … it is something we all just need to eventually do.” Finally, the current e-Business technologies fit poorly for these heterogeneous eBusiness environments. The technologies seem aimed at different kind of networks where partners are more alike both in terms of their implementation capability and eBusiness needs. The current B2Bi frameworks tend to be industry sector standards,

Observations on e-Business Implementation Capabilities

223

yet in these networks cross-industry partnerships are common. This may pose more work on one of the partners, to whom the standard is new or little used. Even when the partners can operate with an “industry native” standard, they still need to agree on a large number of details pertaining to the documents being exchanged, and this raises the bar for engaging in such transactions with partners with whom only a limited number of transactions is expected. There have also emerged e-Business development needs for these networks that are entirely beyond the objectives of currently-available tools. “[Even when we need not map between standards], so we can use papiNet, the papiNet document, it contains a few mandatory fields and then a large number of optional fields that have to be nonetheless agreed on, whether the field is used and how it is used and so forth and this has to be done always with every partner... it is a standard but it is not, like, throughout in that sense.” “It is not realistic with SME's. On the other hand, the profile of our SME and larger suppliers is so different, the volumes on SME's are small and their business is more like services. So there has perhaps not been a business case for us to invest resources into developing B2Bi with these partners.” “We've been thinking about an electronic procurement system, but … we are missing a tool/technology for managing the product data as a universally accessible catalogue or similar [e-Business service]. [Therefore] We cannot move on with this concept.”

5

Discussion and Conclusions

Our research objective was to investigate and characterize the issues that hinder large traditional non-IT industry organizations in implementing e-Business solutions. Our intent was to expand current research by, first, involving a class of organizations that has previously been little studied, and, secondly, by taking into account the heterogeneous nature of the business environment in which the organizations collaborate. Our findings indicated that the organizations were facing not any single, dominant obstacle but rather a mixture of closely-intertwined, accumulative challenges. Lack of IT resources and expertise had resulted in the organizations' relying increasingly on outsourcing. But this seemed to rather limit than support their ability to take proactive, long-term IT development initiatives such as would be required for implementing e-Business. The organizations were also lacking e-Business knowledge and thus had no conscious, organizationally-agreed e-Business strategies defined. As a result, they were progressing with uncertain, haphazard steps that supported no specific e-Business objectives. A major factor here was that without sufficient resources and knowledge, e-Business development projects, which appear particularly challenging for requirements engineering, had became very demanding to carry out. Finally, the heterogeneous business environment posed additional challenges. The business partners had highly-versatile e-Business implementation capabilities, which

224

T. Pesonen and K. Smolander

impeded e-Business initiatives in the network. The available e-Business technologies also fit poorly for use in this environment. In total, we established 11 factors, which we categorized under three higher level classes – IT resources, e-Business knowledge and the business environment – the first two of which were mainly internal to the organizations, while factors under the business environment were external and not in the organization's immediate control. We developed the factors into a model showing their relationship and the order of severity. In view of earlier e-Business research, we were surprised to find that some eBusiness adoption barriers associated with SME's, such as insufficient resources and knowledge [32, 33, 35], and the influence wielded by a single manager in e-Business decision making [15], could also afflict very large organizations when driven to slim down IT spending to minimum. The organization's size alone will not guarantee eBusiness implementation capability. We consider the issues discovered about the heterogeneous business environment to be very important. The current e-Business technologies, such as B2Bi framework standards, fit poorly for this kind of environment due to the wide variety of eBusiness implementation capabilities found there and the propensity of these networks to span industry sectors. This is not surprising, as, according to our best knowledge, heterogeneous business environments have not been studied previously from e-Business requirements’ point of view. Instead, the B2B interaction environment is often expected to be SME-to-SME oriented or otherwise “flat” and more homogeneous than we found it to be. It is possible that this explains some of the difficulties industry elsewhere, too, has experienced in moving towards deeper business-to-business integration. Our plan is to follow up this research by further exploring the SME-MNC B2B interactions, which pose the most difficult interoperability obstacle. We want to understand the constituents of these B2B scenarios and specify concrete interoperability requirements for more feasible e-Business implementations in the case study networks.

References 1. Kotinurmi, P.: E-Business framework enabled B2B integration. Doctoral Dissertation, Helsinki University of Technology, Department of Computer Science and Engineering, Software Business and Engineering Institute (2007) 2. Karuranga, E., Frayret, J.M., D’Amours, S.: Measurements and Determinants of Supply Chain Collaboration. CIRRELT-2008-17, CIRRELT (2008) 3. UN/EDIFACT, http://www.unece.org/trade/untdid/welcome.htm 4. ASC X12, https://www.x12.org 5. Salminen, A.: EDIFACT for business computers: has it succeeded? Standard View 3, 33–42 (1995) 6. Pesonen, T.: Quality Standards in e-Business Software Development. Master of Science Thesis, Lappeenranta University of Technology (2009)

Observations on e-Business Implementation Capabilities

225

7. Nurmilaakso, J.M.: Adoption of e-business functions and migration from EDI-based to XML-based e-business frameworks in supply chain integration. International Journal of Product Economics 115, 721–733 (2008) 8. Nurmilaakso, J.M., Kotinurmi, P., Laesvuori, H.: XML-based e-Business frameworks and standardization. Computer Standards & Interfaces 28, 585–599 (2006) 9. Medjahev, B., Benatallah, B., Bouguettaya, A., Ngu, A.H.H., Elmagarmid, A.K.: Business-to-business interaction issues and enabling technologies. The International Journal on Very Large Data Bases 12, 59–65 (2003) 10. Alonso, G., Casati, F., Kuno, H., Machiraju, V.: Web Service Concepts, Architectures and Applications. Springer, Berlin (2004) 11. Bih, J.: Service oriented architecture (SOA) a new paradigm to implement dynamic ebusiness solutions. ACM Ubiquity (2006) 12. Gamage, A.H.H.: E-Business Impact on SCM in the apparel industry operating between a developing and a developed economy. Brunel University, Buckinghamshire Chilterns University College, Faculty of Enterprise and Innovation (2007) 13. OECD Information Technology Outlook 2004. OECD (2004) 14. Li, M.S., Cabral, R., Doumeingts, R., Popplewell, K.: Enterprise interoperability research roadmap. Final version (Version 4.0), http://ftp.cordis.europa.eu/pub/ist/docs/directorate_d/ e-Business/ei-roadmap-final_en.pdf 15. Fillis, I., Johannson, U., Wagner, B.: Factors impacting on e-Business adoption and development in the smaller firm. International Journal of Entrepreneurial Behaviour & Research 10, 178–191 (2004) 16. Beheshti, M.H., Salehi-Sangari, E.: The benefits of e-Business adoption: an empirical study of Swedish SMEs. Service Business 1, 233–245 (2007) 17. Ramayah, T., Yan, L.C., Sulaiman, M.: SME e-readiness in Malaysia: Implications for Planning and Implementation. Sasin Journal of Management 11, 103–120 (2005) 18. Anastasiou, M., et al.: ATHENA Interoperability Framework, AIF (2010) 19. Coin IP Enterprise Collaboration & Interoperability, http://www.coin-ip.eu/ 20. CEN workshop Agreement on: Feasibility Study on e-Business Interoeprability Test Bed (GITB). CEN/ISSS Interoperability e-Business Forum. CWA 16093:2010 (E) (2010) 21. Rashid, M.A., Al-Qirim, N.: E-Commerce Technology Adoption Framework by New Zealand Small to Medium Size Enterprises. Research Letter of Information and Mathematical Science 2, 63–70 (2001) 22. Tan, Y.H., Bjorn-Anderson, N., Klein, S., Rukanova, B.: Accelerating Global Supply Chains with IT-Innovation. Springer, Heidelberg (2011) 23. Strauss, A., Corbin, J.: Basics of Qualitative Research: Grounded Theory Procedures and Techniques. SAGE Publications, Newsbury Park (1990) 24. Glissmann, S.: IT Service-Oriented Investment Analysis, A Comparison of In-House versus Software-as-a-Service Deployment Solutions. Dissertation no. 3685, Pro BUSINESS Verlag, Berlin (2010) 25. Schroth, C., Pemptroad, G., Janner, T.: CCTS-based Business Information Modelling for Increasing Cross-Organizational Interoperability. In: Enterprise Interoperability II. New Challenges and Approaches. Springer, Heidelberg (2007) 26. Ling-Hong, T.: Commius architecture (final version). Commius ICT 2007 GA 213876 (2007) 27. Seaman, C.B.: Qualitative Methods in Empirical Studies of Software Engineering. IEEE Transactions on Software Engineering 25, 557–572 (1999) 28. Glaser, B., Strauss, A.: Discovery of Grounded Theory, Aldine, Chicago (1967)

226

T. Pesonen and K. Smolander

29. PEPPOL (Pan-European Public Procurement Online), http://www.peppol.eu/results 30. Barratt, M.: Understanding the meaning of collaboration in the supply chain. Supply Chain Management: An International Journal 9, 30–42 (2004) 31. Interoperability cost analysis in the US automotive supply chain. Planning report 99-1, Maryland: National Institue of Standards and Technology (1999) 32. Abid, A.A., Rahim, M.M., Scheepers, H.: Experienced Benefits and Barriers of e-Business Technology Adoption by SME suppliers. Communications of the IBIMA. J. (2011) 33. Pulli, H.: Factors Affecting the Adoption of E-Invoicing in South Karelia. Master’s Thesis, Lappeenranta University of Technology (2005) 34. Khatibi, A., Thyagarajan, V., Seetharaman, A.: E-commerce in Malaysia: Perceived Benefits and Barriers. Vikalpa 28, 77–82 (2003) 35. Olatokun, W., Kebonye, M.: e-Commerce Technology Adoption by SMEs in Botswana. International Journal of Emerging Technologies and Society 8, 42–56 (2010) 36. da Silva, M.M.: Challenges for EDI Adoption by Small and Medium-size Enterprises. In: dos Reis, A.P., Isaías, P. (eds.) IADIS International Conference e-Society 2003, pp. 603–607 (2003) 37. Andersen, V.K., Henriksen, H.Z.: B2B E-commerce Diffusion: The Efficacy of Institutional Discourse. In: Doukidis, G., Mylonopoulos, N., Pouloudi, N. (eds.) Social and Economic Transformation in the Digital Era. Idea Group Publishing, Hershey (2003) 38. Brunnermeier, S.B., Martin, S.A.: Interoperability costs in the US Automotive Supply Chain. Supply Chain Management 7, 71–82 (2002) 39. Legner, C., Lebreton, B.: Business Interoperability Research – Present Achievements and Upcoming Challenges. Electronic Markets 17, 176–186 (2007) 40. Cartwright, J.: RosettaNet Automated Enablement. In: Material Composition Workshop, RosettaNet, August 30 (2004) 41. ITAIDE Project, http://www.itaide.org 42. GENESIS Project, http://genesis-fp7.eu 43. Smolander, K., Rossi, M.: Conflicts, Compromises, and Political Decisions. In: Siau, K., Erickson, J. (eds.) Principle Advancements in Database Management Technologies: New Applications and Frameworks, pp. 82–104. IGI Global (2010) 44. Ross, W.J., Weill, P., Robertson, D.C.: Enterprise Architecture as Strategy; Creating a Foundation for Business Execution. Harvard Business School Publishing, Boston (2006) 45. Macher, J.T., Mowery, D.C., Simcoe, T.S.: E-business and disintegration of the semiconductor industry value chain. Industry and Innovation 9, 155–181 (2002) 46. Daniel, E., Grimshaw, D.J.: An Exploratory Comparison of Electronic Commerce Adoption in Large and Small Enterprises. Journal of Information Technology 17, 133–147 (2002) 47. Hetemäki, L., Nilsson, S.: Information Technology and the Forest Sector. Final Report by the IUFRO Task Force, IUFRO World Series, vol. 18 (2003) 48. Mustonen, T., Karvonen, M., Soininen, L., Hellsten, K., Kässi, T.: Digitizing Business Processes in the Intersection of Energy, Forest and ICT Industries. Lappeenranta University of Technology, Technology Business Research Center, Research Report 13 (2008) 49. OECD Small and Medium Enterprise Outlook, OECD (2002) 50. Byrne, J., Singh, M., Tian, X.: Does size matter? A study comparing e-business implementation in large and small companies in Australia. In: Karmaker, N., Isaias, P. (eds.) Proceedings of the IADIS International Conference E-Commerce 2005, pp. 251–258 (2005)

Development of e-Services in Latvian Insurance Industry: The Critical Role of Quality Benchmarking Valdis Janovs1 and Vladislav V. Fomin2 1

Turiba School of Business Administration, Latvia [email protected] 2 Vytautas Magnus University, Kaunas, Lithuania, Turiba School of Business Administration, Latvia [email protected]

Abstract. The purpose of this paper is to report on the development of comparable set of metrics, or else – a benchmarking standard – on the Latvian non-life insurance market. In this research we used a case study methodology, deploying a number of research strategies, namely: in-depth interviews with heads of industry associations and a survey of representatives of insurance broker companies. We argue that for such information-intensive service industry as insurance to contribute to the development of e-Society, (e-)service offering must be based on common quality benchmarking standards to allow consumers and entrepreneurs have unbiased information on and be able to compare the offered services from different firms. Mutual interaction though e-services between society, enterprises and insurance industry contribute to the development of e-Society, and such service further can contribute to the improvement of insurance industry. Keywords: benchmarking, quality, criteria of quality, non-life insurance industry.

1

Introduction

In 1999 Council of the European Union announced an ambitious vision of Europeans inhabiting Information Society (also dubbed as “e-Society” or “e-Europe”) by 2005 [9]. More than a decade down the road, we have learned the hard way that the success of establishing Information Society cannot be measured by the availability of (often very complex) services provided to citizens over the Internet. e-Society must be built on the principle of maximally even distribution of knowledge among the citizens and businesses. The ultimate measure for success must be the extent to which people are aware about the availability of relevant content, are using the services, and the percentage of population using the services [10, p.285]. The vision of e-Europe was to bring forth “changes, the most significant since the Industrial Revolution” [9, p.2]. However, the issue of interoperability and compatibility of services remains to be a serious obstacle even in the most highly ranked esocieties in the world [24, p.26]. The challenge of making e-services available, known, and used goes beyond the “technical” aspect of the services’ build-out. The aim of this work is to report on the development of comparable set of metrics, or else – a benchmarking standard – on the Latvian non-life insurance market. T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 227–236, 2011. © IFIP International Federation for Information Processing 2011

228

V. Janovs and V.V. Fomin

Authors argue that for such information-intensive service industry as insurance to contribute to the development of e-Society, (e-) services offering must be based on common quality benchmarking standards to allow consumers and entrepreneurs have access to unbiased information on and be able to compare the offered services from different providers.

2

From e-Services to e-Society

What can be referred to as the first, focused statement on e-Society development was the 1993 Clinton Administration’s policy initiative on National Information Infrastructure (NII) [26]. European Union responded to the NII with what is often referred to as Bangemann Report in 1994 [25]. The principles laid out in these documents set the scope for the development of e-Society: everything that produces, contains, processes, or uses information, in whatever form, or whatever media, as well as the people who develop the information, applications, and services [18, p.163]. Looking at the history of national informatization processes in Europe and elsewhere one can notice that the task of building e-Society was far too often understood (or tackled) as bringing a large number of services online [10], i.e., seeing e-Society as a composite of the largest possible amount of e-Services. The lessons learned from the past, however, tell us that the mere availability of eservices doesn’t inevitably cause the transformation of practices [20, p.8] and least so the economies. The programs aimed at bringing about e-services often fail to contribute to the formation of effective e-Society due to cultural, political or other reasons [19]. In the context of our work, one key distinction has to be made in the e-service development – that between availability of information and knowledge. Knowledge about content, quality, pricing of services, as opposed to information about services, facilitates consumers’ weighed decision-making, whereas abundance of information about existence, availability of disparate e-services leads to confusion and consumers’ inability to make a favorable decision. Even distribution of knowledge among the citizens and businesses reduces knowledge asymmetry and facilitates market and democratic processes [5], thus bringing about the sought for transformation of society. Excess of information leads to mental stress [13]. Today, insurance services play a major role in Europe’s economic growth and development, generating premium income of over €€ 1’100bn and investing almost €€ 7’500bn in the economy [6]. Hardly any insurer in Europe does not present information about offered services on the Web or allow the citizens making transactions online – i.e., offering e-services. However, to date, there are no unified non-life insurance quality benchmarking standards in Europe. Such standards would allow reduce the amount of information consumers are exposed to and serve in establishing grounded knowledge on the (comparable) quality of available services. Existence of quality standards for (insurance) services would allow communicating information on available services in uniform and understandable format, thus helping individuals and enterprises make weighed purchase decisions.

Development of e-Services in Latvian Insurance Industry

229

In Latvia, efforts to set a common quality benchmarking standard in non-life insurance market were launched already in 2005 by LIBA – insurance brokers’ association of Latvia. The first author participated in these quality standard-setting efforts as one of the members of LIBA. Since then, Latvian insurers ranking is developed and maintained by LIBA. Six years down the road, neither those insurance quality benchmarking activities, nor the insurers ranking itself are widely known outside LIBA. Knowledge about benchmarking possibilities and results, although available on the Web, is not distributed beyond the narrow circle of insurance professionals. Such situation presents yet another example, where availability of e-services does not contribute to the formation of e-Society.

3

The Role of Standards in the Formation of e-Society

Infrastructures evolve from different and relatively independent from each other technologies and practices that are meshed into a single overarching structure [11]. Successful informatization projects must bring about novel socio-technical configurations, which link together regulatory framework, communications infrastructure, user practices, etc. [15, p.1257]. Standards become crucial elements in informatization processes because of their ability to coordinate activities between and within diverse social groups [21]. Traditionally, the role of standards in informatization projects was studied from technological viewpoint – standards enable interconnectivity, compatibility and interoperability of multiple technologies [16]. However, for the proliferation of e-services to lead to e-Society, informatization process must successfully align diverse interests of participating groups: government organizations, entrepreneurs, consumers, etc. [14]. Standards are also a basis for grouping and comparing diverse and disparate services. Negotiating which technological element or work practice must be chosen over its alternatives to become a part of e-service offering is often complicated due to the possible economic consequences of these decisions. A body or a firm, which succeeds in promoting its favorable technical or service solution as a common standard, often receives large returns, whereas its competitors may be effectively locked out or provided only with residual market niches [23]1. This motivates vendors to adopt protectionist policies for their products, forcing customers into lock-in to a specific product [1]. Such behaviors cause major obstacles in the creation of e-Society as they act counter to even knowledge distribution. Both for emerging technology infrastructure and for emerging e-service platform, standards are both necessary and helpful in that they early on limit the technical or service design space and help obtain a sufficiently fast implementation of a working

1

Sometimes this “battle of the systems” can culminate with the invention of devices that make possible the interconnection between incompatible systems [17] T.P. Hughes, "The Evolution of Large Technological Systems," in Wiebe E. Bijker, Thomas P. Hughes, and Trevor J. Pinch, ed., The social construction of technological systems: New directions in the sociology and history of technology, Cambridge: MIT Press, 1993, pp. 51-82.

230

V. Janovs and V.V. Fomin

design with a large enough user base. This is critical for the emerging markets, where chaotic competition needs to be organized relatively quickly around a relatively stable set of concepts [12]. Standards thus help reduce the risk of choosing underperforming services or products among entrepreneurs as well as consumers. Finally, for e-service offering to contribute to the development of e-Society, the service must contribute to the improvement of economy. Standard-based services help coordinate and organize service offering in such a way, which enables benchmarking and control over distance [7, 22, p.89, 27] – i.e., service providers, their customers and other stakeholders can communicate meaningful knowledge on service offering between one another over the Internet. 3.1

Standards and Benchmarking

Benchmarking assists businesses in identifying potential targets for improvement. As a systematic process for improving performance, benchmarking has gained a great popularity worldwide since the 1980s. As a classic of benchmarking, Camp observed, “establishing operating targets based on the best possible (industry) practices is a critical component in the success of every business” [4]. These “best industry practices” can be dubbed as “quality of operations”, if the quality is defined as one of the core aspects of organizational competiveness. In general, organizational operational quality is aggregate of internal and external practices, aimed at satisfying customers, catering to the needs of society, management, and the expectations of shareholders and employees. Thus, benchmarking moves management thinking from a purely internal focus on organizational modus operandi to one that is external and competitive and can lead to revolutionary rather than evolutionary change [3]. The benchmarking process is usually defined to include four parts, often undertaken continuously or through numerous iterations [8]: 1) analyze the position you are currently in; 2) find someone who is performing measurably better; 3) learn from them what they are doing to achieve that performance; and 4) adapt your practices and processes as a result of that learning and thus implement relevant changes which will effect superior performance in your organization. Thus, the benchmarking process starts by asking a question “what to analyze and how to evaluate”? Looking at the benchmarking process through quality lens, it is important to conduct identification and comparison of quality criteria to obtain an understanding on how to determine which practices are achieving superior quality performance levels. By benchmarking key activities and processes of different firms operating in the same business domain, one can demonstrate given company’s effectiveness in comparison to other service providers, convincing external stakeholders that the company management remain in the best mode of delivery of the service in question [2]. Surprisingly enough, insurance companies in Latvia do not carry out mutual quality benchmarking, and as a result of that neither consumers, nor the insurance companies could rank their service offering against one another in terms of quality. This, in turn, does not allow effective decision making, leads to confusion and, hence, forms a serious obstacle on the way to formation of effective e-services market in Latvia.

Development of e-Services in Latvian Insurance Industry

4

Case Study: Non-life Insurance Industry in Latvia

4.1

Research Design

231

In this research we used a case study methodology to investigate whether there are any (common) standards for benchmarking in insurance industry in Latvia. In our case study we deployed a number of research strategies, namely: in-depth interviews with heads of industry associations and a survey of representatives of insurance broker companies. Three open-ended in-depth interviews were conducted with principals of the three Latvian associations representing insurance industry – LIBA, LIA and LPBA (description of these associations is provided further in the text). Interview with the head of LIBA lasted for 106 min., with the head of LIA – for 90 min, with the head of LPBA – for 56 min. All interviews were audio-recorded, notes of interviews were written down. To compliment interview data, results of insurance broker companies’ survey, conducted by Latvian insurance brokers association (LIBA) were corroborated. Corroboration of data from the interviews and survey, complemented and verified by (industry) expert knowledge of the first author and secondary sources contributed to establishing internal validity of the case data. 4.2

Non-life Insurance Industry in Latvia

There are 10 non-life insurance companies operating in Latvia. All of them are offering different kind of e-services online. Simplest form of e-services is possibility to communicate via their web page using chat function or VOIP – most common one in use is Skype. Most common e-services are those of selling insurance online. As an informational tool, online calculators for determining product/service’s price are used. Car, travel, personal accident and private property insurance are common type of eservices offered and sold through Internet. Several technologically advanced insurers offer additional e-services such as: possibility to submit insurance claim and further monitor claim handling process, log in clients to their personal accounts and overview information about active insurance policies. Several insurance broker companies are offering even more sophisticated eservices to customers. Brokers are using different information and communication technologies to acquire information about insurance price offers using price calculators available on different insurers’ web sites, combining results in one offer and displaying it for clients in their web pages. One of the most popular aggregate e-services of an insurance broker is to obtain/calculate, compare prices and sell Motor Third Party Liability (MTPL) insurance to car owners. Here, the only criteria for comparison is price. For MTPL insurance such benchmarking is reasonable, because MTPL policy conditions are defined by law. However, in the case when other kind of insurance is to be offered, a substantial (level of) benchmarking of available on the Latvian market products and insurance services’ quality must be undertaken, but neither the process nor the results can be easily communicated to the consumer. Peculiar enough, while insurance companies in Latvia do not benchmark their performance against one another, benchmarking of insurers in Latvia has been done since

232

V. Janovs and V.V. Fomin

2007 by insurance brokers’ association LIBA. The developed benchmarking reflect criteria which are specifically important from brokers point of view – insurers may have different opinion what should and should not be benchmarked. LIBA’s benchmarking is based on the results of annual survey of insurance companies titled “Insurers’ performance evaluation by insurance brokers.” All members of LIBA participate in this survey evaluating each insurer’s performance, according to 9 criteria, by giving marks from 1 to 5, where 1 is the lowest possible performance evaluation, and 5 is the highest evaluation. According to this evaluation benchmarking is performed and the ranking of the insurers is made, insurance brokers are licensed, they are experienced insurance professionals, so they can be considered as insurance field experts. Some insurers include the results of the survey in their annual reports, which also show validity of the survey. One insurer even boosted on own competitive superiority having scored the highest in this survey. Insurance industry in Latvia is represented by three associations. Insurance companies are represented by Latvian Insurers Association (LIA), which works in Latvia since 1993. LIA unites 18 insurance companies and branches of foreign insurers (10 non-life and 8 life), which control approximately 99.8% of the total Latvian insurance market. LIA represents the common interests of the insurance industry of Latvia. The association discusses issues that are relevant to the insurers, as well as informs the society about topics significant for the clients. LIA members offer to their clients all kinds of insurance including motor, property, health, life, as well as pension and savings insurance. Latvian Insurance Brokers Association (LIBA) was founded in 2000 by insurance broker companies registered in the Republic of Latvia. Currently there are 104 insurance broker companies in Latvia, 45 of them are members of LIBA. The aim of the LIBA is to develop insurance brokers’ market, raise the quality of insurances services and representation of common interest of the members. The other organization representing Latvian insurance brokers is Latvian Professional Brokers Association (LPIBA). LPIBA was founded in 2000 by insurance broker companies registered in the Republic of Latvia, but distinctively from LIBA members of LPIBA are mainly foreign capital companies. Currently there are 6 insurance broker companies in Latvia who are members of LPIBA. The summary of the opinion expressed in the in-depth interviews by the three heads of associations representing insurance industry enabled formulating the quality benchmarking criteria. The first association, LIA, does not currently conduct any quality benchmarking activities, but they do admit the necessity for such benchmarking. The head of LIA determined such quality criteria: • Client satisfaction (client references, loyalty); • Client complaints (the number of complaints submitted to LIA ombudsman, Finance and Capital Supervision Commission – state governed regulatory institution in Latvia, as well as the number of complaints in media); • Claim handling agility and attitude (quickness, simplicity and accessibility of the process of handling claims); • Reputation (evaluation of the insurer by customers, partners, media and other institutions);

Development of e-Services in Latvian Insurance Industry

233

• Accessibility (number of affiliates, 24 hour call centers, accessibility through internet); • Service level (kindness of employees, IT service level); • Concessionality (interpretation of insurance conditions in clients’ favor). All aforementioned criteria are focused on a client, all aimed to client’s satisfaction. In addition to those criteria there exists another perspective. As described by the head of LIA: There are 2 groups of insurers in Latvia. The first group is insurers with Western European owners, and the second is domestically owned companies. There is an assumption that Western European companies have better quality practices than domestically owned ones. Therefore it can be concluded, that ownership of the company, can be indirectly determined as a quality criteria. The other association – LIBA – has been completing insurer’s quality benchmarking already for 3 years for its own purpose. The interview with the head of LIBA revealed that: • There is an organization of insurance industry professionals apart from LIA, who have independent opinion about insurers; • LIBA independently has developed they own insurers quality criteria – shown in survey. • Those criteria are specifically defined for insurers evaluation from brokers’ point of view. • LIBA criteria differ from LIA criteria because they are defined for different evaluation purposes. In order to summarize their members’ opinion on the quality of Latvia’s insurance companies LIBA has conducted a survey to benchmark insurance companies according to the following criteria: • Performance agility (how quickly insurer serves clients, brokers, claims); • The quality of insurers’ product and services (coverage, deductibles, exceptions, obligatory conditions, conditions of compensation, other conditions); • Price level (comparison of prices to the similar products from competitors); • Insurers’ public reputation (how clients evaluate insurer, reviews from clients); • Is insurer well known (do clients recognize particular insurer, evaluation of insurers’ marketing activity); • Insurers’ attitude towards insurance brokers (insurers’ employees attitude -friendly, arrogant, other and public expressions about brokers); • Do insurers treat direct clients, and broker clients equally (do broker clients receive equal offer in the sense of price and service); • How quickly insurance claims are handled; • Does insurer compete fair (Does insurer try to cheat broker by addressing client directly). As it can be concluded from the surveys’ questions, LIBA benchmarks insurers from their stakeholders’ perspective – as partners/distributors of insurance companies, and simultaneously as representatives of their clients.

234

V. Janovs and V.V. Fomin

It is noteworthy that LIBA considers more important to benchmark their partners – insurers – than to benchmark themselves. Such position can be explained by the fact that insurers are “owners” of the product (i.e., services provided to the customers), and in the sense of quality they are more influential than insurance brokers, which are “distributors”.

5

Analysis of the Case Study Results – Benchmarking Standard as a Basis for e-Services

As admitted by the principals of the examined associations, there is no methodological and comprehensive quality evaluation approach implemented for benchmarking non-life insurance services in Latvia. There is a need (and potential benefits) for an adequate benchmarking standard to be established. There is no quality benchmarking conducted between members of associations. However, insurance association members are benchmarked by members of another insurance market association – insurance brokers association. Since brokers are participants of the same insurance market, and work in close cooperation with insurers, they claim to be experts in the field. According to the research it can be concluded that companies in Latvia’s insurance industry have an understanding about quality benchmarking, but they do not conduct benchmarking according to any model or methodology. Further research is needed to evaluate different models and to possibly choose the best one for the insurance industry. The authors suggest to the associations in the industry to conduct educational explanatory work aimed at convincing companies to participate in benchmarking, to conduct comparison of quality criteria in order to ascertain position organization is currently in, and which practices are achieving superior performance levels. We suggest that if industry representatives (associations) themselves could achieve mutual consensus about quality standards, and establish mutually accepted qualitative benchmarking criteria and standards within insurance industry, then insurance industry could overcome the gap from almost non-existing publicly accessible insurers qualitative benchmarking to effective e-service offering. In this case standards are needed to build clearly defined and trustful environment firstly in industry itself, and afterwards between industry and society. Specifically, we argue that for the insurance industry’s e-service development to contribute to the broader development of eSociety, the following has to be done: 1. Knowledge must be accessible to the society in a simple and understandable form, such as ranking of insurers. At same time, criteria and ranking formation principles must be clear and easily accessible for anyone interested. Currently this is not the case. 2. Results of and knowledge about insurers qualitative benchmarking must be communicated throughout all stakeholders’ e-resources, thereby ensuring wider information dissemination and maximally even distribution of knowledge among the citizens and businesses. 3. Stakeholders involved in benchmarking process must be clearly identified, since different stakeholders have different criteria and understanding about quality, and

Development of e-Services in Latvian Insurance Industry

235

insures brokers evaluation reflect reality from brokers as stakeholders point of view. Such detailed explanation can prevent misperception of rankings as it can currently happen with LIBA ranking. 4. Insurance industry stakeholders such as insurers themselves, banks, leasing companies, supervising government institution and clients are not currently involved in quality benchmarking and have potential to participate in this process if credible and efficient tools are provided by industry. Internet can be used as a platform for collaborative effort of the stakeholders. 5. Since e-environment is agile and variable, it requires constant effort from industry to keep information on offered services up to date, thus safeguarding its credibility. Timely information updates seem to be a challenge for the industry, as can be judged by widespread presence of outdated information on websites of organizations we have studied. 6. Inasmuch as successful positioning and promotion of e-services contributes to the prevalence of knowledge, the knowledge leads to formation of opinion on competing services in the society. Proper Internet-based feedback mechanisms can help improve the original e-service offering if consumers’ opinions are fed back to the insurance industry. 7. The interaction that can take place around e-services between society, enterprises and insurance industry would be a great example of “the spirit of e-Society” and would contribute to the improvement of the industry and economy, too.

References 1. Arthur, W.B.: Competing Technologies, Increasing Returns, and Lock-In by Historical Events. The Economic Journal 99, 116–131 (1989) 2. Bowerman, M., Francis, G., Ball, A., Fry, J.: The evolution of benchmarking in UK local authorities. Benchmarking: An International Journal 9(5), 429–449 (2002) 3. Boxwell, R.J.: Benchmarking for Competitive Advantage. McGraw-Hill, New York (1994) 4. Camp, R.C.: Benchmarking: The search for industry best practices that lead to superior performance. American Society for Quality Control, Quality Press, Milwaukee, WI (1989) 5. Castells, M.: The Rise of the Network Society, 2nd edn. (2000); The Information Age: Economy, Society and Culture, vol. I. Blackwell Publishers, Ltd., Oxford (1996) 6. CEA, About CEA - the European insurance and reinsurance federation. CEA - European insurance and reinsurance federation, Brussels (2011) 7. Ciborra, C.U., Braa, K., Cordella, A., Dahlbom, B., Failla, A., Hanseth, O., Hepso, V., Ljungberg, J., Monteiro, E., Simon, K.A. (eds.): From Control to Drift. The Dynamics of Corporate Information Infrastructures. Oxford University Press, New York (2001) 8. Codling, S.: Benchmarking. Gower, Brookfield (1998) 9. Council of the European Union, eEurope. An Information Society for All (1999) 10. Daniel, E.M., Wilson, H.N.: Dynamic capabilities in e-business transformation. European Journal of Information Systems 12, 282–296 (2003) 11. Edwards, P.N.: Y2K: Millennial Reflections on Computers as Infrastructure. History & Technology 15, 7–29 (1998) 12. Edwards, P.N.: Infrastructure and Modernity: Force, Time, and Social Organization in the History of Sociotechnical Systems. In: Misa, T.J., Brey, P., Feenberg, A. (eds.) Modernity and Technology, pp. 185–226. MIT Press, Cambridge (2003)

236

V. Janovs and V.V. Fomin

13. Farhoomand, A.F., Drury, D.H.: Managerial information overload. Commun. ACM 45(10), 127–131 (2002) 14. Fomin, V.V., Lyytinen, K.: How to distribute a cake before cutting it into pieces: Alice in Wonderland or radio engineers’ gang in the Nordic Countries? In: Jakobs, K. (ed.) Information Technology Standards and Standardization: A Global Perspective, pp. 222–239. Idea Group Publishing, Hershey (2000) 15. Geels, F.W.: Technological transitions as evolutionary reconfiguration process: a multilevel perspective and a case-study. Research Policy 31, 1257–1274 (2002) 16. Hanseth, O., Monteiro, E.: Inscribing behavior in information infrastructure standards. Accounting, Management and Information Technologies 7(4), 183–211 (1997) 17. Hughes, T.P.: The Evolution of Large Technological Systems. In: Bijker, W.E., Hughes, T.P., Pinch, T.J. (eds.) The Social Construction of Technological Systems: New Directions in the Sociology and History of Technology, pp. 51–82. MIT Press, Cambridge (1993) 18. Kahin, B.: The U.S. National Information Infrastructure Initiative: The Market, the Net, and the Virtual Project. In: Kahin, B., Wilson, E. (eds.) National Information Infrastructure Initiatives: Vision and Policy Design, pp. 150–189. MIT Press, Cambridge (1997) 19. Kazama, N.: The regional informatization policies of the local governments in Japan: Why they fail to... Information Infrastructure & Policy 5(4), 279–290 (1996); Presents information on Japan pertaining to the national government and local governments, which have promoted policies to utilize information technology in regional areas. Evaluation of regional informatization policies; How reasons for failure is analyzed; Information on the policy process of local governments; Details of future prospects 20. Kraemer, K.L., King, J.L.: Information Technology and Administrative Reform: Will the Time After E-Government Be Different? CRITO Research Report, University of California, Irvine (August 2003) 21. Lyytinen, K., Fomin, V.V.: Achieving high momentum in the evolution of wireless infrastructures: the battle over the 1G solutions. Telecommunications Policy 26(3-4), 149–170 (2002) 22. Rolland, K.H., Monteiro, E.: Balancing the Local and the Global in Infrastructural Information Systems. The Information Society 18(2), 87–100 (2002) 23. Schilling, M.A.: Technological lockout: An integrative model of the economic and strategic factors driving technological success and failure. Academy of Management Review 23(2), 267–284 (1998) 24. The Danish Government, Using IT Wisely: IT and Telecommunications Policy Action Plan, Report, Ministry of Science, Technology, and Innovation (October 2003) 25. The High-Level Group on the Information Society, Europe and the global information society: Recommendations to the European Council, Council of the European Union (1994) 26. The White House, The Administration’s Agenda for Action, NII (1993) 27. Yates, J.: Control through communication: The rise of system in America. John Hopkins University Press, Baltimore (1989)

A New Model to Study ICT Adoption in Connection with Physical Activity – The TAMPA-Model Eija Koskivaara1, Raija Laukkanen2, and Olli J. Heinonen3 1

University of Turku, Information Systems Science, Rehtorinpellonkatu 3, FIN-20500 Turku, Finland [email protected] 2 Polar Electro Oy, Professorintie 5, FIN-90440 Kempele, Finland and Department of Health Sciences University of Oulu, Oulu, Finland [email protected] 3 Paavo Nurmi Centre and Department of Health and Physical Activity University of Turku, Kiinamyllynkatu 10, FIN-20520 Turku, Finland [email protected]

Abstract. Physical inactivity is one of the leading risks for mortality worldwide. One of future main drivers for physical activity could be information and communication technology (ICT) gadgets and services that support active way of living. This study is the first attempt to model ICT adaption and awareness matrix to motivate and use ICT to support daily physical activity. This is important for individuals and also for information society and healthcare organizations to cope with physical inactivity challenge. They urgently require new approaches and to reshape their customers’ inactive lifestyle. End-user experience data for this research was collected with semi-structured questionnaire from subjects who used pedometers for a three-month- period. In the questionnaire there were questions related to motivation, awareness and also use of information technology connected to physical activity. Based on the results, usage habits and patterns of pedometers, this paper classifies the end-users into four distinctive groups. This classification is called TAMPA-model (Technology Adaption Matrix for Physical Activity -model). Keywords: ICT supported physical activity, pedometer, end-user experience, TAMPA-model.

1

Introduction

A sedentary and unfit lifestyle increases the risk of many chronic diseases and conditions and even decreases longevity [1]. In addition, physical inactivity is globally the fourth highest leading risk for mortality [2]. Physical inactivity and sedentary way of life are modern rising risk factors and they occur worldwide in high, as well as in low income countries. Physical activity (PA) has many scientifically proven health enhancing effects and PA is also effective in preventing and treating specifically lifestyle connected T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 237–244, 2011. © IFIP International Federation for Information Processing 2011

238

E. Koskivaara, R. Laukkanen, and O.J. Heinonen

diseases. Interestingly studies also confirm that long-term physical activity e.g. brisk walking is associated with significantly better cognitive function and reduced risk of dementia [3, 4]. In many countries PA has been promoted by recommendations. These recommendations vary from country to country. The recommendations for youth, adults and older adults in USA are probably the most well-known [5]. The Finnish recommendations for PA are in line with these recommendations: For substantial health benefits, adults should accumulate weekly at least two hours and thirty minutes of moderateintensity exercise, or one hour and fifteen minutes vigorous physical activity. Aerobic activity should be performed in episodes of at least 10-15 minutes. Activity should be spread throughout the week. Adults should also do muscle-strengthening activities (eight to 10 strength-training exercises, eight to 12 repetitions of each exercise) twice a week. Despite many benefits of the PA, initiation and maintenance rates of PA in the general population have been rather disappointing [6]. The sedentary lifestyle is a growing phenomenon of our modern society. The latest Finnish survey reveals that less than half of Finns fulfill physical activity recommendations [7]. Increasing PA requires a change in people’s everyday lifestyle and habits. Indeed, a newly published data in Finland indicate that a physical fitness is an emerging factor of social divide in Finland [8]. The data shows that people who perform some kind of PA at least three times a week are more satisfied with their well-being and social life than those who engage PA less than once a week. To become physically more active is challenging as today there are several new attractive leisure time competitors like social media, computers and console games. As new gadgets and innovations are penetrating the society into deeper levels, the need for studying their usefulness to increase PA and wellness becomes imperative. PA is a widely studied in medicine and sports. More than two hundred high impact studies report new technology used to improve encouragement to achieving recommended PA levels [9]. To really increase PA in different populations we have to understand how normal subjects use information and communication technology (ICT) to support their PA. Technology acceptance and the intention to use information systems are widely studied in information systems since the technology acceptance model (TAM) was published [10]. Usefulness and ease of use are the key factors of acceptance of technology. Individual adoption of innovation like ICT gadgets and services (pedometers, accelerometers, mobile in-built accelerometers, heart rate monitors, social networking, sport gaming and computer based counseling systems, global positioning technologies, and mobile entertainment electronics) can be divided into 5 categories. The categories are: innovators, early adopters, early majority, late majority and laggards [11, 12]. This segmentation is called the DOI-model (Diffusion of Innovation). It seems that pedometers are already at least in the “early majority” -phase and therefore can reveal more on ICT adaption in PA. User acceptance of IT has been thoroughly studied in UTAUT-model (Unified Theory of Acceptance and Use of Technology) [13]. Performance expectancy, effort expectancy, social influence, and facilitating conditions are direct determinants of usage intention and behavior; Gender, age,

A New Model to Study ICT Adoption in Connection with Physical Activity

239

experience, and voluntariness of use moderate the impact of the four key constructs on usage intention and behavior [13]. In chapter two we discuss ICT devices that can be used to support PA with special focus on pedometer. The study design and data analysis are described in the next chapters. The paper ends in presenting the Technology Adaption Matrix for Physical Activity (TAMPA).

2

Use of Pedometers for Physical Activity in Information Society

ICT has penetrated into the society to a level where it has started to show collective impact on the physical and mental state of people throughout the world. This challenges us to reshape our habits. This requires also society and healthcare organizations to reshape their structure and function. Technology along with effective decision making combining motivational and environmental factors may definitely improve the health level of the people. Future important drivers for increasing PA will be ICT gadgets and services like pedometer. A pedometer is a step counting device, usually portable and electronic or electromechanical. It counts each step taken by movement detection. The advantage of pedometer is their low cost and easy use. More advanced pedometers can even estimate e.g. walking distance and burned daily calories. Pedometers can have a motivation impact for subjects wanting to increase PA. Even a target of daily 10 000 steps is universally known originating from Japan in the early 90’s [14]. A more detailed step data for PA estimation is [15]: • Less than 5000 daily steps suggest physical inactivity. • 5.000-7.499 daily steps are typical for of normal daily activity excluding sports or exercise. • 7.500-9.999 steps likely include some exercise or additional walking and/or a job that requires more walking and thereby classify individuals somewhat active. • 10.000-12.000 daily steps express a physically active way of living. • Individuals who take more than 12.500 daily steps are physically highly active. The daily steps have been studied [9, 16].

3

Research Design

This study aims to model ICT gadgets and services adaption and awareness matrix to motivate and use ICT to support daily physical activity. This model developing was carried out in a Step-Shape project which aims to study end-user experience of ICT gadgets and services that support active way of living. The data collection is described in detail on the recent article on Step-Shape project where the focus was on pedometers increasing PA [17].

240

E. Koskivaara, R. Laukkanen, and O.J. Heinonen

Briefly, the subjects were recruited via K5 www-page during January - February 2010. K5 is a special Fun Run event for females and it has also over twenty year’s tradition of promoting PA on local level. Participants of the study may download a special sheet to report daily steps with instructions. The subjects were asked to report their daily steps and the sheet itself counted the daily mean, weekly mean, weekend mean, monthly mean, maximal daily steps and the cumulative step number. The sheet also calculated the change of direction in weekly total steps as well as the absolute weekly change in total steps. The step data collection occurred during three months from February to April. Simultaneously a semi-structured self-reported questionnaire on pedometer use was used. The questionnaire was team designed together with experts in physical activity, sports and exercise medicine and information systems. The information systems questions were influenced on TAM-, DOI- and UTAUT-models. The questionnaire was pilot-tested in a small group of physically active females. After the pilot test some questions were reshaped and few questions were added. The questionnaire included 26 questions; seven were focused on the use of pedometer, twelve questions were related to motivational, environmental and actual PA change issues. The questionnaire contained also four open and feedback comments. The remaining three questions focused on different kind of step results achieved during the project. At the end of step collection period the participants could open the web-questionnaire from the same page where they had downloaded the step-table. The open comments were analyzed by the main author with special focus on the interrelations of ICT and PA. The numerical data was analyzed by the Webrobol-program (http://w3.webropol.com).

4

Data Analysis

Altogether 66 persons opened the questionnaire. Finally ten persons participated this study. Nine of them were women. Mean age was 42 years. All the participants used their own Omron walking style II pedometer (Omron Healthcare Co., Ltd., Kyoto, Japan). On average they had 3.4 years pedometer use experience. The participants reported that pedometer was user-friendly and easy to use. Instructions to use the pedometer had been clear. The pedometer was like a “part of the body”. The biggest challenge was to remember to take it with in the morning. Nine participants fulfilled the step table with computer; two of them did it on daily basis, seven at least once a week. The computers were either own computers or workplace computers. Five out of ten subjects reported that they fulfilled the PA recommendations before start while during the Step-Shape project nine out of ten fulfilled the recommendations. The mean step number showed an increasing trend during the project (Table 1). The user experience issues were organized to 1) ease of use and useful issues (questions 1-6) 2) neighborhood and traffic issues (#7-10) 3) motivational issues (#11-19) 4) lifestyle issues (#20-21) and 5) attitude issues (#22-26) (Table 2). User experience was reported with the following options: 1 = strongly agree, 2 = agree, 3 = no effect, 4 = disagree, 5 = strongly disagree.

A New Model to Study ICT Adoption in Connection with Physical Activity

241

Table 1. Steps taken during the 12 week Step-Shape project Mon-Fri daily mean (12 wk) Sat-Sun daily mean (12 wk) Daily mean steps February daily median March daily median April daily median Maximal individual daily steps Median daily steps

Step count 8879 7964 8326 8570 7964 8493 28.000 7575

Table 2. Self-reported user experience (as mean value) during the 3 month Step-Shape project # 1 2 3 4 5 6

Value 1.1 1.2 1.3 1.3 1.4 1.8

7 8 9 10

1.6 1.4 1.2 2.1

11 12 13

1.6 1.7 2.2

14

2.4

15 16 17 18 19

2.1 2.6 2.4 2.4 2.4

20 21

2.4 2.4

22 23 24 25 26

3.6 4.4 1.6 1.8 2.3

Ease of use and useful The pedometer was easy to use. The step-table was easy to use. The use of pedometer did not require any special thinking. The use of the step-table did not require any special thinking. The pedometer was useful. The step-table was useful. Neighbourhood and traffic I prefer PA in sunny weather. I am engaged to PA in my neighbourhood in the evening. I am engaged to PA in my neighbourhood at daylight. Traffic in the neighbourhood is safe. Motivation The use of pedometer motivated me walking more often. I wanted to know the amount of my daily steps. The use of the step-table motivated me to walk more often. Other reasons e.g. I have activated my family members. I am going to continue step collection, this in fun. I take stairs always as possible. The use of pedometer helped to go for a walk. The use of the step-table helped to go for a walk. I wanted to change my lifestyle. I wanted to become physically more fit for the summer. I want to lose some body weight. Lifestyle I feel physically more fit. I am physically more active. Attitude The Step-Shape project was unfair because some activities do not provide steps. The Step-Shape project was boring. The Step-Shape project was interesting. The Step-Shape project motivated and inspired me. The Step-Shape project was innovative.

242

E. Koskivaara, R. Laukkanen, and O.J. Heinonen

Ease of use got the highest values among users. All in all, in 24 out of 26 questions the response was classified as positive (mean value < 3).

1 26

5

2

25

3 4

24

4

3

23

5

2

22

6

1 21

7 0 8

20 19

9 18

10 17

11 16

12 15

13 14

Fig. 1. Item radars for different end-user issues (1-6: ease of use and useful issues; 7-10: neighbourhood and traffic issues; 11-19: motivational issues; 20-21: lifestyle issues; and 22-26: attitude issues)

The following issues were taken up in the open comments: Step-Shape project really supported and motivated my PA. I wished to lose some weight, but I did not succeed. However, I feel more compact and my physical fitness is better than before the project. I feel myself younger. Step-shape project really motivated me walking. Stepshape project motivated me to competing with myself. I recognized how inactive I really am. And it is very little. I was glad to notice that my daily steps are easily 5000, I am normally physically very active person. Already my normal day includes pretty many steps. Without pedometer this positive activity would be a little bit unclear. Only keeping a pedometer make me more physically active. Eight person reported to continue monitoring their daily steps with pedometer after the Step-Shape –project.

5

TAMPA-Model

After analyzing the user experience we aimed to modeling classification for ICT adoption and PA habits. The level of PA, awareness and use of ICT were selected as major determinants for the model.

A New Model to Study ICT Adoption in Connection with Physical Activity

243

Four main categories were formed (Fig. 2.): 1) Subjects who are highly engaged with ICT are ICT -actives (ICT-Act). Their interest in PA is low. 2) Subjects who are physically very active are PA-actives (PA-Act). They have some awareness of ICT supported PA but they do not use it actively. Their experience may also be negative towards ICT supported PA. 3) Individuals who are active both in technology and exercise are Double Actives (D-Act). They know and use technology in many ways, also to support their PA. They meet and often exceed the PA recommendations. 4) Double Inactives (D-Inact) are physically inactive. Their awareness of ICT supported PA is minimal or does not exist. This categorization is called the TAMPA model (Technology Adaption Matrix for Physical Activity).

Fig. 2. TAMPA -model (Technology Adaption Matrix for Physical Activity. (ICT-Act = ICT – actives, D-Act = double actives, PA-Act = PA –actives, D-Inact = double inactives).

We suggest that the four different TAMPA segments need different kind of ICT and PA approaches to motivate and inspire subjects increasing their daily physical activity to health enhancing levels. PA promoters can apply this model to identify the level and use of technology for their customers. It is anticipated that more and more individuals are connected to different kind of ICT gadgets and applications. In some cases this kind of connectivity may decrease the physical activity level, which in turn may have deleterious health effects. The graphical demonstration shows what attention is “cost-effective”, whether technology or PA as the primary driver. More subjects are needed to test and validate the TAMPA-model.

244

E. Koskivaara, R. Laukkanen, and O.J. Heinonen

Acknowledgements. Authors like to thank M.Sc. Annika Tasala for co-operations in research setting establishments.

References 1. Blair, S., Haskell, W.: Objectively measured physical activity and mortality in older adults. The Journal of American Medical Association 296(2), 216–218 (2006) 2. Who Global Health Risks. Mortality and burden of disease attributable to selected major risks, WHO Library Cataloguing-in-Publication Data. WHO Press (2009) 3. Abbot, R., White, L., Ross, G., Masaki, K., Curb, J., Petrovitch, H.: Walking and dementia in physically capable elderly men. The Journal of American Medical Association 292(12), 1447–1453 (2004) 4. Weuve, J., et al.: Physical activity, including walking, and cognitive function in older women. The Journal of American Medical Association 292(12), 1454–1461 (2004) 5. Physical Activity Guidelines Advisory Committee. Physical Activity Guidelines Advisory Committee Report. U.S. Department of Health and Human Services, Washington, DC (2008) 6. Estabrooks, P., Glasgow, R., Dzewaltowaki, D.: Physical activity promotion through primary care. The Journal of American Medical Association 289(22), 2913–2916 (2003) 7. UKK. Increasing of the people’s physical activity requires changes in their living (2011), http://www.ukkinstituutti.fi (retrieved April 28, 2011) 8. Redera. Finns 2011 Class division into happy satisfied and passive dissatisfied people is already here (2011), http://www.redera.fi (retrieved April 21, 2011) 9. Tudor-Locke, C., Hart, T.L., Washington, T.L.: Expected values for pedometer-determined physical activity in older populations. International Journal of Behavioral Nutrition and Physical Activity 6, 59 (2009) 10. Davis, F.D.: Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly 13(3), 319–339 (1989) 11. Rogers, E.M.: Diffusion of Innovations, 4th edn. Free Press, New York (1995) 12. Rogers, E.M.: Diffusion of Innovations. Free Press, Glencoe (1962) ISBN 0612628434 13. Venkatesh, V., Morris, M.G., Davis, G.B., Davis, F.D.: User acceptance of information technology: Toward a unified view. MIS Quarterly 27(3), 425–478 (2003) 14. Hatano, Y.: Use of the pedometer for promoting daily walking exercise. Journal of the International Committee on Health, Physical Education and Recreation 29, 4–8 (1993) 15. Tudor-Locke, C., Bassett Jr., D.R.: How Many Steps/Day Are Enough?: Preliminary Pedometer Indices for Public Health. Sport Medicine 34(1), 1–8 (2004) 16. Hirvensalo, M., Telama, R., Tammelin, T., Yang, X., Viikari, J., Raitakari, O.: Finnish women take more daily steps than Finnish men (in Finnish). Liikunta ja Tiede 47(4), 18–21 (2010) 17. Koskivaara, E., Laukkanen, R., Heinonen, O.J.: ICT supporting daily physical activity – with special reference to pedometers in the Step-Shape project. Scientific Journal of Riga Technical University, a Special Issue of Applied Computer Systems (forthcoming, 2011) ISSN 1407-7493

A New Electronic Path to the Hospital – A Prenatal Observation System Arnim Nethe and Irene Krebs Brandenburg University of Technology, Konrad-Wachsmann-Allee 1, 03046 Cottbus, Germany [email protected]

Abstract. Premature babies represent a particularly endangered group of children. At the same time, their parents are frequently traumatized by the early birth and full of fear for their child. The question which arises here now is how parents can build up a relationship to their child. This article describes a visitor system which establishes the connection between child and parents and, on the other hand, could substantially support the hospital personnel with their daily work. A clinical example for this e-health application is given. This is a faster step in the direction of e-society. Keywords: telemonitoring, IT-system, secure access, virtual visitor system, health care, early born children.

1

Introduction

Telemedical applications, such as intelligent assistance systems and telemonitoring systems, and their technical implementation gain more and more importance. According to BBC Research, the global market for telemedical systems alone will annually grow to approx. 11 billion Euros till 2012 [6]. Based on current research an application for a new target group in telemedicine is presented in this paper. For this practice-oriented example only few field tests have been conducted worldwide [4, 7].

2

Premature Infants and Ill Newborn Children

Premature infants represent a particularly endangered group of children. At the same time their parents are frequently traumatized by the early birth and full of fear for their child. In this situation we must not leave them alone, but must help them to be able to be there for their children because especially in the first weeks of life they particularly need the attention of their parents. About 60,000 premature babies are born in Germany per year - with rising tendency: Despite a sinking birth rate premature babies constitute an ever larger share of all newborn children. Meanwhile 8.5% of all newborn children are premature babies. Thus they form the largest children’s group of patients in the Federal T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 245–254, 2011. © IFIP International Federation for Information Processing 2011

246

A. Nethe and I. Krebs

Republic. Of the 60,000 premature children per year about 8,000 children are born with a birth weight of less than 1,500g. Despite the early birth a large part of these children is in a good course of development. About 30% are risk patients with different forms of restriction and handicap. In addition to this, investigations verify that the social environment is of particular importance for the optimal support of premature and ill newborn children. After early birth, parents experience a phase of severe stress, accompanied by fears, feelings of guilt and helplessness. Frequently the situation asks too much of them. It is difficult for them to establish a relationship to their child in the hospital. Incubators, medical equipment, cannulae and tubes are additional barriers. Particularly very premature children spend a long time in the incubator in a nonnatural environment in the neonatological intensive care unit. Studies indicate that the attachment behavior of premature infants frequently exhibits uncertainties or is even disorganized, as it is often observed with abused children. Therefore a failed establishment of relationships can result in permanent impairments of the cognitive, social and emotional development of the child. The fear for the survival of the child, the concern about the health development and possible impairments as well as the own feelings of guilt of having “caused” the situation impair the basis for a stable parenthood [1].

3

A Concept for the Improvement of the Situation

The question which arises here now is how a mutual bond can be established and developed. In addition, the psycho-social support of the families is a significant factor which must be considered. A possible solution can be a virtual visitor system in the neonatology intensive care unit. If parents could see the baby again the evening before bedtime, that would minimize the fears and increase the bond to the child. Thus family life can also be more relaxed since the new family member can be integrated into the family. The parents can show the child to friends and relatives and thereby integrate him/her into the family environment. In addition, the virtual visitor system would supplement the control services of the hospital personnel. It is possible for the personnel to additionally observe the child by a camera from the monitoring room. In cases when the child accidentally pulls out the tubes, the nursing personnel can react faster because this can be noticed by the live monitoring. Thus, on the one hand, the virtual visitor system would strengthen the mutual bond between child and parents and, on the other hand, would substantially support the hospital personnel in their daily work. Thus the nurses, hospital aides and physicians can do their jobs more productively and more efficiently. The approach of installing a prenatal monitoring system - we call it “BabyLive” in hospitals cannot be taken for granted today, although the advantages are obvious and a technical implementation would be possible without any problems. In the following part it is explained how the approach described can be implemented and which factors have to be considered.

A New Electronic Path to the Hospital – A Prenatal Observation System

4

247

IT Concept for the Implementation of the BabyLive System

In an IT concept all technical prerequisites for the implementation of the project in a hospital are described. A basic prerequisite for the IT application is, among other things, the development and implementation of a uniform IT security concept with the goal of guaranteeing the integrity, confidentiality and availability of data and services, the operation of an efficient, highly-available, preferably open and safe clinical center net for the transmission of digital data of all kinds as well as the build-up and operation of a need-oriented support structure. In addition to the basic equipment of the workplaces, the users expect that a number of superordinate IT services are provided with high availability and reliability. This includes, among other things, a central listing service for purposes of user, assigning the parts, the rights management, a network server for the supply of the personal computers, secured access to the clinical center net, IT security checks, as well as reactive and proactive IT safety precautions (e.g. firewall, intrusion detection, intrusion prevention, virus defense, spam defense), email services as well as videoconference services (video transmission of the child’s photographs). 4.1

Services in the IT Environment

In connection with the basic equipment and IT services mentioned above there are also further requirements. They can roughly be divided into four groups: • A platform for rapid informing about current events on the homepage of the clinical center and information about the further development of BabyLive by means of newsletter. • A consultation for informing parents about the use of the service and a catalog for frequently asked questions (FAQ). • Training on general IT topics for nursing personnel and parents for using the service BabyLive. • Technical support for the staff, analysis of the use of the service, assessment and/or planning of IT procurements as well as the procurement of software on the basis of valid contracts. 4.2

Security in the IT Environment

The planning phase covers the threat analysis, a risk analysis, the preparation of a safety strategy and a safety model. In the threat analysis possible dangers are determined and their causes such as organizational, technical or user-related causes are found out. These tasks can only be mastered if certain knowledge and experience in the field of IT security is available and the weaknesses of the existing systems and services are known (see Fig. 1). In the risk analysis the determined threats are evaluated. The evaluation contains, on the one hand, the estimation of the probability of the occurrence of the threats, and,

248

A. Nethe and I. Krebs

Fig. 1. Process model for the implementation of the BabyLive system [5] on the other hand, the estimation of the caused potential damage when a threat occurs. Due to the threat and risk analysis it can be decided which safety requirements are to be set for the system to be designed. These requirements are summarized in a safety strategy. To meet the security needs general basic safety functions can be used. These functions are: • Identification and authentification: Subjects and objects must be clearly identifiably, for example with the access to resources or to certain files. To facilitate identification characteristics for an authentification must be specified, such as passwords and a digital fingerprint. • Rights administration: This concerns the prevention of unauthorized access to objects. Administrative rights must be specified for the objects to be protected, such as reading and writing rights. • Rights examination: In order to implement rights administration the accesses must be controlled. This is the task of rights examination. For this the time and the kind of control must be specified. Further the measures to be taken in cases of unauthorized access must be regulated. • Preservation of evidence: The preservation of evidence is to secure proof so that accomplished actions cannot to be denied later. For this it must be specified which events and information are to be logged and who possesses access to the logged information. • Reprocessing: To retain information privacy shared operational funds must be settled before another process accesses them.

5

Practical Example in the Pediatric Clinic of the Carl-Thiem-Klinikums (CTK) Cottbus

The neonatology unit of the pediatric clinic of the Carl-Thiem-Klinikum Cottbus is one of four institutions of care level 1 in the Federal State of Brandenburg. Care level

A New Electronic Path to the Hospital – A Prenatal Observation System

249

1 means that only these institutions are allowed to bill health insurance companies for the treatment of newborn children under a birth weight of 1250 grams. Further hospitals with neonatology units are located at a great geographical distance from Cottbus. The neonatology unit in Cottbus is divided into two sections. In the intensive room for premature infants in the delivery room ill premature and newborn infants are primarily treated, for example those who need breathing support. The second section is the intensive care unit for premature infants, where those newborn children are cared for who do not need intensive medical treatment (any more). [2] In all areas of the neonatology unit where premature infants are cared for, connections for hospital-internal communication are available. Furthermore, the monitors (security camera screens), which for example display the heart frequencies of the children, are all interconnected with one another. These data can be retrieved by means of administrative software at several terminals at the unit. The clinical center has its own homepage with a babysite, where only photos of newborn children have been made available so far. This site could be supplemented by an extension for retrieving video transmissions of the children. Due to the special operation of a hospital information system it is necessary to parallel implement the complete infrastructure for the BabyLive system.

Fig. 2. Participants and use relationships

After the introduction to IT security it is now to be illustrated how internal and external security can be implemented using the pediatric clinic in the CTK as an example. For the illustration of the section which is to be considered here see figure 2. The internal view represents the Carl-Thiem-Klinikum Cottbus and the participants who use video transmission and can be observed by the camera. The pictures are transferred to the Internet platform. The users of the platform, such as the parents, can then access the Internet platform and the picture transmissions. The Internet platform

250

A. Nethe and I. Krebs

is an interface between the clinical center and the platform users. Additionally to the users already specified, the nursing personnel can also use the service. The nursing personnel can track the picture transmission of all cameras on the computer in the monitoring room. 5.1

Internal Security

For internal security no information should get out which is not intended for the outside world. This could be, for example, the treatment of the child which is not to be filmed and thus must not become visible by the users of the platform of the virtual visitor system, but also the protection against an internal access to the system by unauthorized people. First, possible risks for the clinical center and its personnel must be listed. These are: • The treatment and care of the child by the nursing personnel and the physicians are filmed and transmitted. • Staff of the clinical center could overhear the data traffic. • A coworker can use the service of the video transmission for which s/he does not possess authorization. • By pretending identity a coworker reroutes the data. From the threat analysis the following protection goals result: • Video transmission must be prevented during the treatment and nursing of the child. • The access to the service, the servers and the data must be controlled and checked. Before using the service the personnel must authorize themselves. So that the nursing personnel and the physicians are not observed during the treatment of the child, the camera must either be turned off every time or the bed is put away from the camera. However, this is too difficult, costs time and especially in emergencies cannot always be implemented. Instead a sign could be attached in front of the lens of the camera, for example with the text “I am off” or “I am being fed”, which is placed in front of it every time the child is not to be observed. The solution is simple, does not cost too much time and the user of the Internet platform knows what is happening in these moments. So that the data traffic cannot be overheard by staff, it must be ensured that the net of the video transmission is operated separately from the remaining clinical center net. Thus only the staff of the neonatology unit can access and use the service of the virtual visitor system. Further, authorization before using the service can guarantee that no unauthorized persons have access to the data communication on the net and thus cannot reroute or abuse the data. 5.2

External Security

The ways from the Internet platform to the user of the platform and from the user to the platform are of importance for external security. Possible dangers could be:

A New Electronic Path to the Hospital – A Prenatal Observation System

• • • • • • •

251

Unauthorized access to the Internet platform, Abuse of the access to data, Spoofing of emails by an aggressor, Unauthorized reading of emails, Attack on communication between the interface and the PC of the user, Spoofing of the identity, Execution of malware on the web server by the user.

From the threat analysis the following protection goals result: • • • •

Access to the service and the data must be controlled and checked. Before using the service the user must authenticate himself/herself. The integrity, authenticity and privacy of emails must be guaranteed. Access of the user to the clinical center’s internal net must be impossible.

5.3

Technical Realization

For the implementation of the schematic representation of the BabyLive system shown in Figure 3 the following steps are necessary: On the Internet site of the CarlThiem-Klinikum Cottbus a link is to be established, which is to refer to the site of the BabyLive system. The link indicates the IP address of the SDSL modem of the unit. For safety reasons, an inquiry about the user name and the password will be carried out here. This authentification is global, i.e. each user of the camera platform has this registration for the front page of the cameras. The protection is organized so that no unauthorized persons can access this homepage. Furthermore, this global registration

Fig. 3. Schematic representation of the BabyLive system

252

A. Nethe and I. Krebs

is logged in a log file. This file can also be used for statistical purposes in order to determine how successfully this service is used and if further investments would be worthwhile. A camera is assigned to every child. The parents select the camera which was assigned to them. Again there will be an inquiry about the user name and the password. The router implements port forwarding to the server in the nurses’ room, on which the hosting of the webpage of the camera entrances is run. The IP addresses of the cameras are mapped on the individual IP of the server by the port forwarding, so that the internal infrastructure remains transparent for the user. After the authentification the video transmission of the child becomes visible by means of a live stream point to point. This means that the video transmission is converted into a video/audio stream in real time by means of encoder software. In this way the stream can be transmitted via the Internet. Point to point streaming means that a central server provides a connection between the participants (child/parents). This data exchange is server independent. The video transmission is only stopped by logout or the end of the visiting time. The visiting time is dependent on the organization of the care of the children, as it is also usual with normal visits in the hospital. Furthermore, the visibility of the child in the net can be disturbed by certain events, for example when the child is being fed or examined. As already mentioned, in such events posters are placed in front of the camera.

Fig. 4. Homepage of the Internet platform: “Welcome to BabyLive: With the project BabyLive parents of premature infants at the unit K3 of the hospital for child and adolescent medicine of the CTK are offered the possibility to visit their child "virtually". This means that a web camera is installed at the bed of the child via which parents can see the child at all times of the day and night. Using the field "Kamerazugang" you can select the camera which was assigned to you. Then you only have to authentificate yourselves and see your child live”.

A New Electronic Path to the Hospital – A Prenatal Observation System

253

The relatives take over a significant part of the implementation. They must make themselves familiar with the technology, receive instructions and give their agreement for the video transmission. In order to make the project a success, parents would have to accept and use this hospital service. The extent to which the service offered is used helps decide whether there should be more investment in this project. The fear that the pictorial material could be abused by unauthorized people can be taken away by entrance identification. By this identification only authorized users of the respective camera can see the child. If the relatives forget the access code, the staff of the hospital can help by generating a new access code and a new password (see Fig. 4 and Fig. 5).

Fig. 5. Camera site of the Internet platform

6

Conclusions

Today there is a highly complex IT structure in every hospital. The implementation and adaption of a completely new access path with previously non-existent users for a prenatal observation system increases the complexity of this IT structure. Thus, among other things, new requirements arise regarding the development and implementation of a uniform IT security concept with the aim of guaranteeing the integrity, confidentiality and availability of data and services, organizing the operation of a highly efficient and available, preferably open and secure hospital network for the transmission of digital data of all kinds as well as organizing the setup and operation of a demand-oriented support structure. In addition, there is the training of all people involved on how to use the prenatal observation system.

254

A. Nethe and I. Krebs

A priori a modern hospital is a part of e-health. Thus with this prenatal observation system a telemedical platform is created for the computer affine generation of young parents with which a children’s hospital can significantly improve e-service and in this way can make a big step in the direction of e-society.

References 1. Kinderhilfe, D.: Das Frühförderprogramm der Deutschen Kinderhilfe (FFP), Deutsche Kinderhilfe e.V. (2007) 2. Erler, T., Ellsäßer, G.: Die Gesichter der Opfer - Auswirkungen von Gewalt bei Kindern und Jugendlichen erkennen. Pädiat. Prax 75, S.387–S.395 (2010) 3. Krebs, I., Nethe, A.: Zugang für Alle - Barrierefreies Leben in einer digitalen Welt für eine langfristige Selbstbestimmtheit. In: Demographischer Wandel - Assistenzsysteme aus der Forschung in den Markt (AAL 2011) (2011) 4. LeDuc, K., Haley-Andrews, S., Rannie, M.: An observation unit in a pediatric emergency department: One children’s hospital’s experience. Journal of Emergency Nursing 28(5), 407–413 (2002) 5. Poguntke, W.: Basiswissen IT-Sicherheit. Das wichtigste für den Schutz von Systemen & Daten, 2. Auflage, Witten (2010) 6. Reinhardt, P.: Potenziale für intelligente Assistenzsysteme. DeviceMed - Fachmagazin für Hersteller medizinischer Produkte 9. Jg., 5 (September 2010) 7. Vocera Communications, Inc.: Vocera Benefits Study at The Children’s Hospital at Westmead, Prepared by The NTF Group March 2007 Vocera Communications (2007)

Students’ e-Learning System Usage Outcomes: A Study with a Learning Management System A.K.M. Najmul Islam Turku Center for Computer Science (TUCS), University of Turku, Finland [email protected]

Abstract. E-learning systems such as the learning management systems are widely used in higher education. However, much of the research on e-learning systems focuses on the technology and is limited to the adoption and utilization e-learning systems. In order to develop the potential advantages of e-learning systems, research that addresses the outcomes of the adoption and use of elearning system is needed. This paper proposes a research model for assessing the possible outcomes of learning management systems adoption and use. It was tested by university students (n = 249) participating in hybrid courses using partial least squares (PLS) analysis. Its findings suggest that beliefs about perceived usefulness and perceived ease of use, and how an e-learning system is used assist students’ perceived learning and perceived community building. Keywords: e-learning system use, learning management system, technology acceptance model, use outcomes.

1

Introduction

One of the most significant developments in the use of information technology in universities in the last decade has been the adoption and use of e-learning systems to support the processes of teaching and learning. The learning management system (LMS), also known as course management system (CMS) or the virtual learning environment (VLE), is an e-learning system that has been widely adopted by universities. The LMS is web-based software that is used for the delivery, tracking and managing of education and training online. It contains features for distributing courses over the Internet and online collaboration. Since the late 1990s, the utilization of LMSs for online education has steadily increased in higher education. Nowadays, LMSs have become indispensable tools for online education. Whether focusing on distance education or class-room based education, most universities use LMSs to support and improve learning and teaching processes [1]. For example, in 2005, 95% of all higher education institutions in the UK used course management systems [2]. LMSs are usually implemented across an entire university, faculty, or school, and then adopted by educators who use them in a variety of ways to support course management and student learning [3]. The use of the LMSs in university education has made it easy to augment online education with traditional face-to-face classroom instruction. Such a hybrid or mixed delivery approach allows educators to combine T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 255–268, 2011. © IFIP International Federation for Information Processing 2011

256

A.K.M. Najmul Islam

the advantages of online learning with the benefits of face-to-face instruction [4]. A mix of face-to-face (somewhere between 90% and 10%) and online instruction (somewhere between 10% and 90%) has been argued to be superior to either 100% face-to-face or 100% online courses [31]. However, in order to understand the impact of e-learning systems on students’ learning outcomes and take advantage of the potential of e-learning systems, research that addresses the role of e-learning systems adoption and its use in learning outcomes is needed. In addition, research investigating the factors that influence the adoption and use of e-learning systems is required – if we are to further develop e-learning systems. There has been much prior research focusing on the adoption and use of elearning systems [5; 6]. These studies identified the e-learning system usage intention as the dependent variable and investigated its antecedents or determinants. However, these studies overlooked the outcomes of e-learning system usage. Knowing the outcomes of e-learning system usage is particularly important if we are to evaluate the success of such systems, plan for their future development and achieve better learning outcomes. Thus, this paper investigates the outcomes of e-learning system usage for hybrid courses from the perspective of university students. The paper proceeds as follows. In section 2 we present the literature review and research model. Section 3 is dedicated to the research method and section 4 presents the data analysis results and discussion. Finally, section 5 discusses the implications.

2

Literature Review and Research Model

One of the major focuses of prior e-learning research has been the adoption and continued use of e-learning systems [5; 6]. This research stream has mainly used two theoretical frameworks: the technology acceptance model (TAM) [7] and the expectation-confirmation model (ECM) [8] to investigate individuals’ e-learning system adoption and continued use. These studies mainly investigated the factors that affect the adoption and use of e-learning systems, but they do not consider how these factors, or the use of the e-learning system itself is associated with learning outcomes. Indeed, few studies have gone beyond use to explore the factors associated with learning. McGill and Klobas [3] found that e-learning system utilization influences the perceived impact on learning. Lee and Lee [9] revealed that a number of elearning environment quality related variables affect satisfaction with e-learning. In turn, satisfaction was found to influence academic achievement. Liaw [10] found high correlation between intention to use e-learning and e-learning effectiveness. These studies provide some empirical support about the possible relationships between elearning system use and e-learning system usage outcomes. However, these studies have been conducted with a variety of outcome variables that use different explanatory variables and this has led to models that offer only weak theoretical support. Thus, these studies fall short in explaining the relationship between the antecedents of adoption and use of e-learning systems and their use outcomes, and the relationship between e-learning system use and use outcomes. Hence, a conceptual framework with strong theoretical support is necessary.

Students’ e-Learning System Usage Outcomes

257

Fig. 1 illustrates the conceptual framework of the e-learning system use outcomes.

Beliefs -Perceived ease of use -Perceived usefulness

E-learning use intention/behavior

E-learning use outcome

-Use intention -Use

-Perceived learning assistance -Perceived community building assistance

Fig. 1. Conceptual framework for e-learning system use outcome

The framework is conceptually supported by both the IS adoption model, TAM [7] and the IS success model [11]. TAM presents two behavioral beliefs, perceived usefulness and perceived ease of use as the antecedents of IS usage intention and use. Perceived usefulness is the degree to which an individual believes that the system will enhance his or her job performance and perceived ease of use is defined as the degree to which a person believes that using a system would be free of effort [7]. Following TAM, we posit that these two behavioral beliefs result in e-learning system use. On the other hand, according to the IS success model, IS use brings benefits to both individuals and organizations [11]. Similarly, according to our framework, e-learning system use will positively affect the learning process of individuals. In particular, we argue that two variables regarding individuals’ learning processes, perceived learning assistance and perceived community building assistance, are important. Perceived learning assistance refers to the extent to which the e-learning system assists an individual’s learning while perceived community building assistance refers to the extent to which the e-learning system assists individuals in building a social community. In summary, according to this framework, beliefs may impact on e-learning use outcomes in two ways. First, the beliefs can have a direct impact on e-learning use outcome variables. For example, a useful and/or easy to use e-learning system may assist students by helping them to learn more effectively and by giving them an opportunity to build a social network. Such benefits may help students improve their academic performance. The direct relationship between the beliefs and e-learning use outcomes are supported by a number of prior studies [3; 9]. Second, these beliefs can have an indirect effect on e-learning use outcomes through the mediation of elearning system use. As described earlier, this indirect effect is supported by the IS success model. Based on the conceptual framework, we have developed a research model for evaluating how the use of an e-learning system affects e-learning outcomes. The research model is shown in Fig. 2.

258

A.K.M. Najmul Islam Use antecedents

Use

Use outcomes H6

TAM

H1 H4 Perceived ease of use

Perceived learning assistance

H2

Perceived usefulness

E-learning use H5

H7 H8

H9 Perceived community building assistance H3

Fig. 2. Research model

TAM postulates that perceived usefulness has an impact on IS use. If individuals believe that the e-learning system they use provides them with better information or increased learning control, they are more likely to use the system in the courses for reading and downloading learning materials and interacting with other participants (participating in the discussions, chatting, emailing, etc.). The association between perceived usefulness and e-learning system use is supported by many prior studies [5; 6]. Thus, we make the following hypothesis. H1. Perceived usefulness positively affects students’ e-learning system use. A particular e-learning system can provide value in several ways [12]. First, the system can be more useful when it expands either the quality or quantity of information which the students are able to access. Second, the e-learning system can be useful because it helps the students manage and control their learning process. Prior studies have argued that greater learning control by students leads to better learning outcomes [13; 14], which suggests that a useful e-learning system may increase students’ control of the learning process and eventually affect learning effectiveness. In addition, the e-learning system can be useful because it contains many features that help the participants collaborate with each other. Thus, we propose the following two hypotheses. H2. Perceived usefulness positively affects students’ perceived learning assistance. H3. Perceived usefulness positively affects students’ perceived community building assistance. TAM further postulates that perceived ease of use affects IS use. It has been confirmed by many studies in different contexts including e-learning system utilization [5; 32]. Thus, we propose the following hypothesis. H4. Perceived ease of use positively affects students’ e-learning system use. We further argue that an easy to use e-learning system may assist students’ learning and help to build a collaborative community. The technical design and format will impact on how effectively the participants can share learning materials

Students’ e-Learning System Usage Outcomes

259

and collaborate with each other [36]. In addition, when the students perceive an elearning system as easy to use, they are better able to concentrate on their learning instead of putting effort into learning the system itself. The system provides them with an easy way to collaborate with others, which eases the group work. Following this, we propose the following two hypotheses. H5. Perceived ease of use positively affects students’ perceived learning assistance. H6. Perceived ease of use positively affects students’ perceived community building assistance. Although, students and educators are physically separated in an e-learning environment, it has been argued that students’ engagement is greatly increased in such a learning environment [15]. Increased student engagement can improve learning outcomes, such as problem solving and critical thinking skills [16]. Individuals are assumed to learn better when they discover things by themselves and when they control the pace of learning [17]. This implies that the use of e-learning systems in the courses provides some sort of self-directed learning opportunities which lead to the improvement of learning effectiveness among the students. The studies built upon social network theory argue that students interact more effectively when a social structure enables them to access a larger base of contacts and makes the exchange of information faster [33; 34]. Fast receipt of information may assist learning. For example, educators can upload learning materials in advance so that the students can read them before going to class. This may allow them to learn effectively in classroom situations. In addition, both educators and students can instantly share other course related information by utilizing the database of contacts. They may also use chat or discussion facilities for immediate and fast communication. Thus, it is plausible to assume that the utilization of the e-learning systems provides the opportunity to share information faster. Thus, we propose the following hypothesis. H7. Use of an e-learning system in a course positively affects students’ perceived learning assistance. Studies suggest that an effectively designed and administered online course environment can assist in the creation of a social environment [18]. We further argue that such social interaction will create a social network. For example, face-to-face meetings among educators and students should help to build a supportive social network. Additionally, online activities should provide students with the opportunity to become better acquainted and share their common interests about a particular topic more deeply. It may even be that many students become more active in the online social medium of instruction in comparison with face-to-face situations. In fact, some students may prefer to participate more in online discussions rather than face-to-face. In brief, active online discussions conducted by students may allow them to create a network with both educators and other students. Thus, we propose the following hypothesis. H8. Use of an e-learning system in a course positively affects students’ perceived community building assistance. Lastly, research on the learning processes in face-to-face groups indicates that the development of social community is important for making students feel like insiders in the learning environment, thus contributing to students’ motivation, involvement, and contentment [19]. Following this, we propose the following hypothesis. H9. Perceived community building assistance positively affects students’ perceived learning assistance.

260

A.K.M. Najmul Islam

3

Research Method

3.1

Questionnaire Development

The questionnaire had three parts: demographic questions, questions related to the constructs of the research model, and open ended questions asking to report the students’ satisfaction and dissatisfaction with the target system. Each item corresponding to the constructs has been measured using the seven-point Likert scale, the answer choices range from “Strongly disagree (1)” to “Strongly agree (7)”. Most of these items are adapted from the literature with minor changes in wording reflecting the target system. The measures of perceived usefulness and e-learning system use were adapted from Limayem et al. [22] and Ma et al. [23] respectively. The measures of perceived ease of use were adapted from Hong et al. [24]. The measures of perceived learning assistance were developed from Liaw [10]. Finally, the measures of perceived community building assistance were developed from Paechter et al. [25]. After the questionnaire was drafted, it was first sent to two academic researchers for their review and revised according to their comments and suggestions to make the wording of the items more precise. Then, the questionnaire was sent to 10 students for their review. Overall, the students indicated that the questionnaire was relatively clear and easy to complete. A number of suggestions were made concerning the wording of several items and the overall structure of the questionnaire. The questionnaire was revised according to the given suggestions. In order to avoid the common method bias problem as much as possible, we decided to randomize the questions in the questionnaire during data collection [26]. 3.2

Participants

The target system of this study is the learning management system, Moodle (http://moodle.org/about/). Moodle is an open source course management system and has become very popular among educators for creating dynamic online course websites for students. Moodle can be used to conduct online courses or to augment face-to-face courses (hybrid courses). This study was conducted in an internationally acknowledged, multidisciplinary scientific university in Finland. The university has seven faculties. The university has been using Moodle since 2007 as its platform for creating course pages online. Data was collected via a web-based survey from the students of the university who use Moodle in hybrid courses. A list of students’ email addresses was collected from the Moodle support team in the university. A total of 1100 email invitations were sent to randomly selected students of the university who had been registered in Moodle as student users. One reminder was sent to increase the response rate after a gap of one week. The survey ran for approximately two weeks. After filtering invalid and incomplete responses, we had a total of 249 survey responses that could be used in this study. The demographic information of the respondents is given in Table 1.

Students’ e-Learning System Usage Outcomes

261

Table 1. Demographic information

Gender

Male Female

Frequency 101 148

Percentage 40.56 59.44

Age

less than 21 years 21-30 years 31-40 years >41 years

31 166 25 27

12.45 66.67 10.04 10.84

Experience with the target system

0 – 6 months >6 months – 1 year >1 year – 1 year 6 months >1 year 6 months – 2 years >2 years – 2 years 6 months >2 years 6 months – 3 years >3 years

15 63 10 47 9 27 78

6.02 25.30 4.02 18.88 3.61 10.84 31.33

3.3

Data Analysis

To analyze the quantitative data, we employed partial least squares (PLS) as our analysis approach and utilized the tool smartPLS [27]. PLS is a second generation regression method that combines confirmatory factor analysis with linear regression, and this makes it possible to run the measurement and structural models simultaneously. The qualitative data was content analyzed and used to support the findings of the PLS model. Table 2 shows item wise averages and the loadings of each construct in the model. For each construct the assessment of convergent validity or internal consistency is also included through the composite reliability coefficient [28]. Convergent validity indicates the extent to which the items of a scale that are theoretically related are also related in reality. As we can see from Table 2, all items have significant path loadings greater than the threshold 0.7 recommended Fornell and Larcker [28]. All the constructs have composite reliability values that exceed the threshold recommended by Nunnally [29]. Testing for discriminant validity involves checking whether the items measure the construct in question or other (related) constructs. Discriminant validity was verified with both correlation analysis and factor analysis as recommended by Gefen and Straub [30]. First, the inspection of discriminant validity among variables is based on the correlation between variables and the square root of their respective average variance extracted [28]. As Table 3 shows, the square root of the average variance extracted value for the variables is consistently greater than the off-diagonal correlation values, suggesting satisfactory discriminant validity between the variables. Second, from Table 4 we see that all items have cross loadings coefficients lower than the factor loading on their respective assigned latent variable, suggesting that discriminant validity on the item level is met for all the constructs.

262

A.K.M. Najmul Islam Table 2. Construct items, means and internal consistencies

Construct Perceived usefulness (CR = 0.91; AVE = 0.76)

Item PU-1: Using Moodle is of benefit to me PU-2: The advantages of Moodle outweigh the disadvantages PU-3: Overall, using Moodle is advantageous

Mean 5.33 5.37

std 1.46 1.38

Loading 0.87* 0.91*

5.32

1.71

0.84*

Perceived ease of use (CR = 0.93; AVE = 0.77)

PEOU-1: My interaction with Moodle is clear and understandable PEOU-2: Interacting with Moodle does not require a lot of mental effort PEOU-3: I find Moodle to be easy to use PEOU-4: I find it easy to get Moodle to do what I want to do

4.90

1.39

0.86*

5.15

1.51

0.87*

5.16 4.76

1.54 1.51

0.90* 0.88*

5.67

1.38

0.81*

4.84

1.37

0.93*

4.76 4.31

1.46 1.54

0.92* 0.83*

Perceived learning assistance (CR = 0.93; AVE = 0.76)

PLA-1: Moodle provides flexibility of learning with regard to time and place PLA-2: Moodle assists learning performance PLA-3: Moodle assists learning efficiency PLA-4: Moodle assists learning motivation

PCB-1: Moodle provides opportunities to 3.97 1.76 Perceived establish personal contact with teachers community PCB-2: Moodle makes it easy to do group 3.88 1.57 building assistance (CR work = 0.89; AVE = PCB-3: Moodle provides opportunities to 3.35 1.57 0.72) establish new contacts with other students USE-1: I use Moodle frequently in this 4.82 1.65 E-learning system use (CR academic period = 0.96; AVE = USE-2: I use Moodle heavily during my 4.20 1.58 0.92) study Note: Composite reliability (CR), Average Variance Extracted (AVE), *p < 0.001

0.82* 0.88* 0.85* 0.96* 0.96*

Table 3. Correlation between the variables and squared root of average variance extracted PLA PEOU PU PCB USE PLA 0.87 0.57 PEOU 0.88 0.68 0.65 PU 0.87 0.60 0.36 0.45 PCB 0.85 0.60 0.60 0.70 0.40 USE 0.96 Note: Perceived learning assistance (PLA), Perceived ease of use (PEOU), Perceived usefulness (PU), Perceived community building assistance (PCB).

Students’ e-Learning System Usage Outcomes

263

Table 4. Factor analysis results PLA PEOU PU PCB USE 0.558073 0.648265 0.477433 0.537861 PLA-1 0.811703 0.521947 0.594350 0.519992 0.529865 PLA-2 0.929225 0.501885 0.615953 0.478445 0.521858 PLA-3 0.920721 0.427086 0.520408 0.597071 0.509509 PLA-4 0.827863 0.526706 0.619708 0.375455 0.567165 PEOU-1 0.860389 0.420433 0.503426 0.230550 0.446315 PEOU-2 0.867054 0.501233 0.582400 0.295034 0.543351 PEOU-3 0.896624 0.546963 0.569808 0.355136 0.518067 PEOU-4 0.876356 0.629898 0.577446 0.413334 0.624178 PU-1 0.869696 0.591721 0.602227 0.402224 0.645901 PU-2 0.910331 0.556769 0.531375 0.351760 0.570731 PU-3 0.838663 0.440082 0.212962 0.359335 0.343828 PCB-1 0.819031 0.581021 0.415652 0.448503 0.379125 PCB-2 0.875997 0.480850 0.274310 0.317311 0.300786 PCB-3 0.852910 0.577935 0.563957 0.672974 0.380150 USE-1 0.956699 0.571949 0.578858 0.673771 0.392885 USE-2 0.957184 Note: Perceived learning assistance (PLA), Perceived ease of use (PEOU), Perceived usefulness (PU), Perceived community building assistance (PCB).

4

Results and Discussions

The test of the structural model includes estimates of the path coefficients, which indicates the strengths of the relationships between the dependent and independent variables, and the R-square values, which represent the amount of variance explained by the independent variables. Fig. 3 shows the results of the test of the hypothesized model. Use antecedents

Use

Use outcomes

0.15*

TAM Perceived usefulness

Perceived ease of use

0.33*** 2

0.55*** R = 0.53 E-learning 0.14* use 0.15* 0.24**

0.09ns

Perceived learning assistance 0.34*** Perceived community building assistance 0.28**

*p < 0.05; **p < 0.01; ***p < 0.001; ns: non-significant

Fig. 3. PLS Model results

R2 = 0.60

R2 = 0.22

264

A.K.M. Najmul Islam

As hypothesized, perceived usefulness (β = 0.55; p < 0.001) and perceived ease of use (β = 0.24; p < 0.01) have a significant impact on e-learning use explaining its 53% variance. On the other hand, perceived usefulness (β = 0.33; p < 0.001), e-learning use (β = 0.14; p < 0.05), perceived ease of use (β = 0.15; p < 0.05), and perceived community building assistance (β = 0.34; p < 0.001) have significant effects on perceived learning assistance explaining its 60% variance. Perceived usefulness (β = 0.28; p < 0.01) and e-learning use (β = 0.15; p < 0.05) were also found to have a significant impact on perceived community building assistance. The relationships of the TAM model were supported in our study. As expected, perceived usefulness affects e-learning use significantly. This finding is in line with the findings of [8]. The qualitative data also supported this finding. For example, one student wrote the following: “….Finding all course materials in one place is great….” In addition, we found that perceived ease of use also predicts e-learning use significantly. However, its predictive strength is weaker than perceived usefulness. This finding is in line with those of IS adoption studies [7]. The other hypotheses are discussed below. First, our study results revealed that perceived usefulness strongly impacts on both perceived learning assistance and perceived community building assistance. These relationships were not tested empirically in the prior studies. However, the findings of prior studies provide indirect empirical support for our findings. Johnson et al. [12] conceptualized course performance, course satisfaction and course instrumentality as the measures of elearning effectiveness. They concluded that perceived usefulness has a significant impact on both course performance and course satisfaction. Our finding is also indirectly supported by media synchronicity theory [35]. Media synchronicity theory emphasizes three key features of a media that enable effective communication: parallelism, reprocessibility and rehearsability [35]. Parallelism refers to the medium’s capability to support multiple and simultaneous conversations. Reprocessibility refers to the medium’s capability to support the reexamining and revisiting of a message. Finally, rehearsability refers to the medium’s capability of supporting reediting and refining a message before sending it out. The LMSs provide such facilities, of which the group discussions facility in the e-learning systems is an excellent example. Additional examples include the ability to have multiple conversations, documenting conversation history for future reference, and editing a message before sending it. Thus, our finding about the association between perceived usefulness and perceived learning assistance is in line with the prior literature. The qualitative data also provides support for our findings. For example, one student wrote the following: “….For one course, we had good group discussion on Moodle. The topic was good, and all the students participated and Moodle worked well which made me inspired by the course and assignment in general….” In addition, Dourish, and Bly [37] have examined how synchronous and asynchronous computer mediated communication (CMC) tools deliver and support social awareness. They found the successful delivery of social awareness via the

Students’ e-Learning System Usage Outcomes

265

CMC tools resulted in a sense of community through the maintaining of working relationships and informal communication. Thus, our finding about the association between perceived usefulness and perceived community building assistance is in line with prior findings. Second, we found that perceived ease of use was also significant – although very weak – in predicting perceived learning assistance. In addition, we observed that it had a non-significant impact on perceived community building assistance. We argue that the weak and non-significant impacts can be explained based on prior findings. In particular, adoption studies argue that perceived ease of use is weak at predicting the usage intention for experienced users [7]. In our study, the respondents were experienced with the target system (see Table 1). Thus, perceived ease of use did not contribute strongly to predicting perceived learning assistance and perceived community building assistance. Third, we found that e-learning system use also has a significant direct impact (although very weak) on both perceived learning assistance and perceived community building assistance. These causal relationships have not been tested empirically in prior literature, although there is ample evidence of support for this finding. For example, it is often argued that replacing some of the in-class activities with elearning resulted in higher learning outcomes when compared to traditional face-toface learning. Empirically, Liaw [10] found a high correlation between e-learning utilization intention and e-learning effectiveness. In addition, McGill and Klobas [3] found LMS use positively affects students’ perceived learning impact. On the other hand, several studies found that students are inclined to explore new social ties and links when they participate in Computer Supported Collaborative Learning (CSCL) environment [33]. Thus, we conclude that our findings are supported by prior findings. However, the interesting point is that we found e-learning use is very weak at predicting both learning assistance and community building assistance. The interpretation of this finding is that the extensive use of e-learning system does not necessarily bring benefits to students. Hence e-learning systems should be used in such a way that can help to learn and increase collaboration. Such effective utilization largely depends on an educator’s capabilities to build an effective learning environment with the students. The qualitative data provides further support to this. For example, one student wrote the following: “….All teachers do not know how to use Moodle, some teachers know but they don’t take the full advantage of Moodle….” Another student wrote the following: “….Moodle sites work best when the teachers put timely and valuable information there….” Another student wrote the following: “….I have the feeling that teachers do not really want us to take full advantage of Moodle. For example, the teachers do not really encourage us to use discussion forums. In addition they do not allow us to put external links to the Moodle page which could be important for learning….”

266

A.K.M. Najmul Islam

Finally, perceived community building assistance impacts on perceived learning assistance. This finding is also in line with prior studies indirectly. For example, a number of studies found that building a sense of community is necessary for successful learning outcomes [19; 31]. It suggests that the use of e-learning systems helps students to build a social network with teachers and other students, which leads to more effective learning.

5

Implications

Our study findings have a major theoretical implication. Our study has found how TAM variables affect e-learning usage outcomes. While prior studies only investigated the adoption and use of e-learning systems, our study went beyond elearning use and provided insights into the outcomes of e-learning system adoption and use. Overall, the study’s findings suggest that an e-learning system may assist students in learning and building a collaborative network, if it is used to augment face-to-face education. Our study findings have practical implications for e-learning system designers, educators, and school management. First, the study found that behavioral beliefs (perceived usefulness and perceived ease of use) have significant impacts on perceived learning assistance and perceived community building assistance, which implies that designers need to develop e-learning systems that are useful and easy to use in order to affect students’ learning outcomes. Developing easy to use and useful systems also leads to more usage of such systems by students. However, educators need to remember that the increased usage of e-learning systems may not lead to more effective learning and better community building. Educators are required to put effort into designing courses utilizing e-learning systems so that the students are able to learn effectively. They should add different functionalities to the course pages and encourage the students to use them. For example, educators may encourage the students to participate in the discussion forums. They may announce some incentive, such as the possibility to earn some bonus points for active participation. Finally, an education institute’s management needs to know how to ensure users’ effective usage of e-learning systems. Improving students’ and educators’ knowledge about the e-learning systems should lead to the effective usage of such systems. If they do not have much knowledge of these systems, they are less likely to use them or they may not be able to gain the full benefits of such systems, which may result in ineffective online collaboration. In turn, ineffective collaboration may negatively impact on student learning. Therefore, schools and universities should provide training for both students and educators on how to use their particular e-learning systems most effectively.

References 1. Dalsgaard, C.: Social software: E-learning beyond learning management systems. European Journal of Open, Distance and E-learning (2006), http://www.eurodl.org/materials/contrib/2006/Christian_Dalsg aard.pdf (retrieved March 26, 2011)

Students’ e-Learning System Usage Outcomes

267

2. Browne, T., Jenkins, M., Walker, R.: A longitudinal perspective regarding the use of VLEs by higher education institutions in the United Kingdom. Interactive Learning Environments 14(2), 177–192 (2006) 3. McGill, T.J., Klobas, J.E.: A task-technology fit view of learning management system impact. Computers & Education 52, 496–508 (2009) 4. Edling, R.J.: Information Technology in the classroom: Experiences and recommendations. Campus Wide Information Systems 17(1), 10–15 (2000) 5. Islam, A.K.M.N.: Understanding continued usage intention in e-learning context. In: 24th Bled eConference AIS e-library, Slovenia (2011) 6. Islam, A.K.M.N., Mäntymäki, M.: Culture and student samples as moderators of continued IT usage: A meta-analysis of IS continuance literature. In: 15th Pacific Asia Conference on Information Systems (PACIS). AIS e-library, Brisbane (2011) 7. Davis, F.D.: Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly 13(3), 319–340 (1989) 8. Bhattacherjee, A.: Understanding Information Systems Continuance: An ExpectationConfirmation Model. MIS Quarterly 25(3), 251–370 (2011) 9. Lee, J.-K., Lee, W.-K.: The relationship of e-Learner’s self-regulatory efficacy and perception of e-Learning environment quality. Computers in Human Behavior 24, 32–47 (2008) 10. Liaw, S.S.: Investigating students’ perceived satisfaction, behavioral intention, and effectiveness of e-learning: A case study of the Blackboard system. Computers & Education 51, 864–873 (2008) 11. DeLone, W.H., McLean, E.R.: Information systems success: The quest for the dependent variable. Information Systems Research 3(1), 60–95 (1992) 12. Johnson, R.D., Hornik, S., Salas, E.: An empirical examination of factors contributing to the creation of successful e-learning environments. International J. Human-Computer Studies 66, 356–369 (2008) 13. Ellerman, H.H., Free, E.L.: A subject-controlled environment for paired associate learning. Journal of Computer-Based Instruction 17, 97–102 (1990) 14. Freitag, E.T., Sullivan, H.J.: Matching learner preference to amount of instruction: an alternative form of learner control. Educational Technology, Research, and Development 43, 5–14 (1995) 15. Northrup, P.: A framework for designing interactivity into Web-based instruction. Education Technology 41(2), 31–39 (2001) 16. Liaw, S.S., Huang, H.M., Chen, G.D.: An activity-theoretical approach to investigate learners’ factors toward e-learning systems. Computers in Human Behavior 23, 1906–1920 (2007) 17. Leinder, D.E., Jarvenpaa, S.: The use of information technology to enhance management school education: a theoretical view. MIS Quarterly 19(3), 265–291 (1995) 18. Wang, M., Kang, M.: Cybergogy for engaged learning: a framework for creating learner engagement through information and communication technology. In: Hung, D., Khine, M.S. (eds.) Engaged Learning and Emerging Technologies, pp. 225–253. Springer, Netherland (2006) 19. Wegerif, R.: The social dimension of asynchronous learning networks. Journal of Asynchronous Learning Networks 2(1), 34–49 (1998) 20. Currin, L.: Feelin’ groovy. eLearn Magazine (2003), http://www.elearnmag.org/subpage.cfm?section=articles&articl e=10-1 (retrieved May 6, 2011)

268

A.K.M. Najmul Islam

21. Gatian, A.W.: Is user satisfaction a valid measure of system effectiveness? Information & Management 26(3), 119–131 (1994) 22. Limayem, M., Hirt, S.G., Cheung, C.M.K.: How Habit Limits The Predictive Power Of Intention: The Case of Information Systems Continuance. MIS Quarterly 31(4), 705–737 (2007) 23. Sun, Y., Bhattacherjee, A., Ma, Q.: Extending technology usage to work settings: The role of perceived work compatibility in ERP implementation. Information & Management 46, 351–356 (2009) 24. Hong, S.-J., Thong, J.Y.L., Tam, K.Y.: Understanding continued information technology usage behavior: A comparison of three models in the context of mobile internet. Decision Support Systems 42(3), 1819–1834 (2006) 25. Paechter, M., Maier, B., Macher, D.: Students’ expectations of, and experiences in elearning: Their relation to learning achievements and course satisfaction. Computers & Education 54, 222–229 (2010) 26. Straub, D.W., Boudreau, M.-C., Gefen, D.: Validation guidelines for IS positivist research. Communications of the Association for Information Systems 13, 380–427 (2004) 27. Ringle, C.M., Wende, S., Will, A.: Smart PLS 2.0 M3, University of Hamburg (2005), http://www.smartpls.de 28. Fornell, C., Larcker, D.F.: Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research 18(1), 39–50 (1981) 29. Nunnally, J.C.: Psychometric theory, 2nd edn. McGraw Hill, New York (1978) 30. Gefen, D., Straub, D.: A practical guide to factorial validity using PLS-Graph: Tutorial and annotated example. The Communications of the Associations for Information Systems 16(5), 91–109 (2005) 31. Woods, R., Jason, D.B., Hopper, D.: Hybrid structures: Faculty use and perception of webbased courseware as a supplement to face-face instruction. The Internet and Higher Education 7, 281–297 (2004) 32. Larsen, T.J., Sorebo, A.M., Sorebo, O.: The role of task-technology fit as users’ motivation to continue information system use. Computers in Human Behavior 25, 778–784 (2009) 33. Cho, H., Gay, G., Davidson, B., Ingraffea, A.: Social networks, communication styles and learning performance in a CSCL community. Computers & Education 49(2), 309–329 (2007) 34. Ortiz, M., Hoyos, J., Lopez, M.: The social networks of academic performance in a student context of poverty in Mexico. Social Networks 26(2), 175–188 (2004) 35. Dennis, A.R., Fuller, R.M., Valacich, S.J.: Rethinking media richness: towards a theory of media synchronicity. MIS Quarterly 32(3), 575–600 (2008) 36. Lim, H., Lee, S.-G., Nam, K.: Validating E-learning factors affecting training effectiveness. International Journal of Information Management 27, 22–35 (2007) 37. Dourish, P., Bly, S.: Portholes: Supporting awareness in a distributed work group. In: Proceeding of ACM Conference on Human Factors in Computing Systems, pp. 541–547. ACM, New York (1992)

Antecedents of Post-adoption Expectations: A Study in the e-Learning Context A.K.M. Najmul Islam Turku Center for Computer Science (TUCS), University of Turku, Finland [email protected]

Abstract. This paper examines factors influencing post-adoption expectations toward an e-learning system. Following the expectation-confirmation based information system continuance theory, we use perceived usefulness as the surrogate of post-adoption expectations. Then, we propose a research model based on the prior literature where we hypothesize a number of factors along with the confirmation of initial expectations affect perceived usefulness. The model is tested among the university educators (n = 175) who use an e-learning system to conduct their teaching. The results suggest that post-adoption expectation is driven by confirmation of initial expectations, perceived control beliefs, and perceived system quality. Subjective norm does not have significant impact on post-adoption expectations. The determinants of perceived usefulness explained 68% of its total variance. Keywords: Expectation-confirmation theory, e-learning, Perceived usefulness, Post-adoption expectations.

1

Introduction

During the recent years, e-learning via the Internet has become so common phenomenon that it has boosted the use of software tools that lent structure and management for online courses. These tools are called Course Management Systems (CMS) which feature educators with easy communication to the students, the ability to track students’ progress, and a mean to presenting content for online courses securely. Such e-learning tools are very attractive to the universities because they have many useful features. However, while evaluating an e-learning system like a CMS, educators’ perspective must be considered. The success of such technology depends on the educators’ willingness of its continued utilization. It is argued in the literature that continued utilization is determined by post-adoption expectation and satisfaction [1]. Despite its importance in explaining continued use, a little effort has been given to identify factors that shape post-adoption expectations. However, it is very important for practitioners to know what shapes post-adoption expectations to plan for interventions that could influence post-adoption satisfaction and continued use. It is even more important for the e-learning service providers as the courses utilizing elearning platforms are still evolving. T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 269–281, 2011. © IFIP International Federation for Information Processing 2011

270

A.K.M. Najmul Islam

Prior research studies have assumed perceived usefulness as the surrogate measure of post-adoption expectations [1]. Perceived usefulness is defined as the degree to which a user believes that using a particular system would enhance his or her job performance [2]. Researchers have found a number of antecedents of perceived usefulness in IS adoption studies such as perceived ease of use, subjective norm, image, job-relevance etc [3]. However, most of these studies do not differentiate between pre-adoption and post-adoption stage. Indeed, some factors have been shown to predict perceived usefulness during post-adoption stage. For example, perceived ease of use has been often found as a significant determinant of perceived usefulness during post-adoption IS use [4]. Given that the psychological motivation behind initial adoption and subsequent use being different [1], further research is necessary to underpin the important factors that shape perceived usefulness during post-adoption stage. Thus, this paper addresses the following research question: what are the antecedents of post-adoption expectations toward an e-learning system? In answering this question, we draw different factors from the IS adoption and post-adoption literature and use expectation-confirmation based IS continuance theory [1] as the foundation theory to build our research model. Then, we test our proposed research model by collecting survey data among university educators who use a popular elearning platform, Moodle (http://moodle.org/) to conduct their teaching. The paper proceeds as follows. In section 2, we present the theoretical background. In section 3 we develop our research model. Section 4 is dedicated to the research method and section 5 presents the data analysis results and discussion. In section 6 we describe implications of our findings for theory and practice. Finally, section 7 concludes the paper.

2

Theoretical Background

The root of post-adoption expectations is the expectation-confirmation theory [5]. This theory hypothesizes that consumers’ level of satisfaction with a product/service determines repurchase intention. In turn, consumer satisfaction is determined by two major constructs: initial expectations (pre-purchase expectations) on a product/ service, and discrepancies between expectations and product/service performance (disconfirmation). According to this theory, buyers first develop expectations about a product/service before purchase. Second, their consumption experiences with it build perceptions about its performance. This leads to the buyer either confirming or disconfirming the pre-purchase expectations, after assessing perceived performance against the earlier frame of reference (pre-purchase expectations). A buyer’s expectations are confirmed when the product/service performs as much as expected; negatively disconfirmed when it performs worse than expected; positively disconfirmed when it performs better than expected [5]. Drawing attention to the substantial difference between initial adoption and continued usage, the information system continuance model was developed from the expectation-confirmation theory and empirically tested in voluntary environment [1].

Antecedents of Post-adoption Expectations: A Study in the e-Learning Context

271

The information system continuance model focuses importance on post-adoption expectations rather than pre-adoption expectations. According to the model, perceived usefulness is the measure of post-adoption expectations, since among the cognitive beliefs in IS adoption and usage, perceived usefulness demonstrated itself to be the most consistent and salient one in determining the user intention over time [1]. The information system continuance model posits that the users after first time acceptance and a period of initial use will form an opinion to which their pre-acceptance expectations are confirmed (Confirmation). From this confirmation, the users form an opinion about the benefits (Perceived usefulness). After a period of time, both confirmation and perceived usefulness are the basis of satisfaction with the IS (Satisfaction). Finally, perceived usefulness and satisfaction impact the users’ willingness to continue the IS (intention). The information system continuance model is shown in Fig. 1.

Post-adoption expectations: perceived usefulness Satisfaction

Continuance Intention

Confirmation

Fig. 1. IS continuance model [1]

3

Research Model

Utilizing the expectation-confirmation based IS continuance theory and other complementary theories, prior studies have found a number of antecedents in addition to confirmation that shape perceived usefulness in different contexts including elearning system utilization. These antecedents fall into three broad categories: system quality related factors, individual’s control beliefs, and social aspects related factors. Perceived ease of use [4], perceived playfulness [6], and perceived task-technology fit/perceived compatibility [7] may fall into system quality category. A few studies used perceived competency that falls into individual’s control belief category [6; 8]. Under social aspects category, the researchers tested the effect of subjective norm on perceived usefulness [9]. In this study, to determine post-adoption expectations, we use these three categorical factors in addition to the confirmation of initial expectations. Fig. 2 shows the research model. In the following, we describe these factors and build our hypotheses. Confirmation is a cognitive belief which is defined as the extent to which users’ expectations of e-learning system use is realized during actual use [1]. The expectation-confirmation based IS continuance theory posit that confirmation of

272

A.K.M. Najmul Islam

expectations has an effect on perceived usefulness. Many studies have found a significant effect of confirmation on perceived usefulness in different contexts including e-learning system [4; 6]. Thus, we make the following hypothesis. H1. Educators’ confirmation of expectations of an e-learning system positively affects their perceived usefulness with the e-learning system. Control beliefs are related to an individual's beliefs about the presence of factors that may facilitate or impede performance of the e-learning system use [10]. We argue such control beliefs would lead educators to realize the usefulness of an e-learning system. After using an e-learning system, the educators develop cognitions about the control beliefs that may facilitate or impede performance of the e-learning system use. Educators’ cognitions about such beliefs would lead them to understand the e-learning system’s capabilities and develop appropriate level of post-adoption expectations during the post-adoption stage. Thus, we propose the following hypothesis. H2. Educators’ perceived control beliefs with an e-learning system positively affect their perceived usefulness with the e-learning system. Based on the prior studies in the IS continuance, we operationalize control beliefs as a second-order construct having two reflective constructs: perceived behavioral control and perceived competency. Perceived behavioral control refers to people’s perception of the ease or difficulty of performing the behavior of interest [10]. Perceived competence refers to being effective in dealing with the e-learning system a user works with [11]. It is one of the three psychological needs those are the basis of self-determination theory [12]. System quality is a general perception about a system in terms of its performance and it can be reflected by various system features. We argue that experience with using the e-learning system to conduct various tasks can produce a better understanding of its characteristics, which can assist the user in forming cognitions about its overall usefulness. We argue that if the users of a system develop cognitions about its perceived system quality, it would, in turn, affect their cognitions about its perceived usefulness. Seddon [13] proposed an association between system quality and perceived usefulness in his modified IS success model. Latter, Lee and Lee [14] found a significant causal relationship between e-learning environment quality and perceived usefulness. Thus, we make the following hypothesis. H3. Educators’ perceived system quality of an e-learning system positively affects their perceived usefulness with the e-learning system. System quality has been operationalized in many different ways in IS literature. In general, the manifest variables of system quality in terms of a web-based system are access convenience, flexibility, integration, response time, sophistication, reliability, accessibility, stability, system speed, usability, ease of use, navigation and network speed [15]. Among these variables compatibility, ease of use, access, and reliability are the most important system quality related variables during post-adoption stage [15]. Thus, we model system quality as a multi-dimensional construct composed of four reflective constructs: perceived compatibility, perceived ease of use, perceived access, and perceived reliability in this paper. Perceived compatibility is defined as the degree to which an e-learning system is perceived as being consistent with

Antecedents of Post-adoption Expectations: A Study in the e-Learning Context

273

existing values, needs and experiences of the users [16]. Perceived ease of use is defined as the degree to which a person believes that using a system would be free of effort [2]. Perceived access refers to the degree of accessibility, responsiveness, stability, and availability of the e-learning system [15]. Perceived reliability refers to the dependability of the e-learning system operation [17]. Social norm refers to the person’s perception that most people who are important to him think he should or should not perform the behavior in question [18]. Social norm has been shown to have a direct impact on behavioral intention in literature [3]. It has also been theorized to have an indirect effect on behavioral intention through perceived usefulness [19]. It implies that when one perceives that an important referent thinks one should use a system, one incorporates the referent’s belief into one’s own belief structures [19]. Thus, we hypothesize the following. H4. Social norm positively affects their perceived usefulness with the e-learning system.

Confirmation

Social norm H1 H4 Perceived usefulness H3 H2 System quality

Control Beliefs

PBC

PC

EOU

Access

REL

COMP

Fig. 2. Research model

4

Research Method

4.1

Questionnaire Development

Each item corresponding to the constructs has been measured using the seven-point Likert scale, with answer choices ranging from “Strongly disagree (1)” to “Strongly

274

A.K.M. Najmul Islam

agree (7)”. Most of these items are adapted from the literature with minor changes in wording reflecting the target technology. The measures of confirmation and perceived usefulness were adapted from Bhattacherjee [1] while the measures of perceived ease of use were adapted from Davis [2]. The measures of perceived compatibility and subjective norm were adapted from Moore and Benbasat [16] and Venkatesh et al. [20] respectively. Finally, the measures of perceived competence were adapted from Sorebo et al. [8]. After the questionnaire was drafted, it was first sent to two academic researchers for their review and revised according to their comments and suggestions to make the wording of the items more precise. Then, the questionnaire was sent to 30 educators at the university, so that they could review it. Overall, the educators indicated that the questionnaire was relatively clear and easy to complete. A number of suggestions were made concerning the wording of several items and the overall structure of the questionnaire. The questionnaire was revised according to the given suggestions. 4.2

Data Collection

Data was collected via a web-based survey from the educators of a university in Finland who use Moodle for their teaching purpose. A list of educators’ email addresses was collected from the Moodle support team in the university. A total of 1012 email invitations were sent to the educators of the university who had been the registered Moodle users. Two reminders were sent to increase the response rate in two weeks gaps. The survey ran for approximately one and half months. After filtering invalid and incomplete responses, we had total 207 survey responses. In this particular study, we were interested about those users who had been conducting at least one course using Moodle at that academic year. After filtering the survey responses according to this criterion, we ended up with 175 usable responses. The demographic information of the respondents is given in Table 1. Table 1. Demographic information

Gender

Male Female

Frequency 74 101

Percent 42.3 57.7

Age

21-40 years >41 years

90 85

51.4 48.6

Experience with the target technology

0 – 18 months >18 – 36 months >36 months

54 83 38

30.9 47.4 21.7

4.3

Data Analysis

We employed partial least squares (PLS) as our analysis approach and utilized the tool smartPLS [21]. PLS is a second generation regression method that combines

Antecedents of Post-adoption Expectations: A Study in the e-Learning Context

275

confirmatory factor analysis with linear regression, and this makes it possible to run the measurement and structural models simultaneously. Table 2 shows item wise averages and loadings of each construct in the model. For each construct the assessment of convergent validity or internal consistency is also included through the composite reliability coefficient [22]. Convergent validity indicates the extent to which the items of a scale that are theoretically related are also related in reality. As we can see from Table 2, all items have significant path loadings greater than threshold 0.7 recommended by Fornell and Larcker [22]. All the constructs have composite reliability values that exceed the threshold recommended by Nunnally [23]. Table 2. Construct items, means and internal consistencies Construct Perceived usefulness (CR = 0.92)

Ease of use (CR = 0.93)

Perceived Compatibility (CR = 0.91)

Confirmation (CR = 0.91)

Item PU-1: Using Moodle increases the control with my teaching plan PU-2: Using Moodle enables me to accomplish teaching information sharing more quickly PU-3: Using Moodle is of benefit to me PU-4: The advantages of Moodle outweigh the disadvantages

Mean 4.59

std 1.33

Loading 0.84*

5.8

1.33

0.78*

5.18 5.31

1.28 1.23

0.85* 0.87*

5.65

1.22

0.80*

4.46

1.52

0.88*

4.38

1.52

0.75*

4.49 4.27

1.52 1.61

0.94* 0.91*

COMP-1: Using Moodle fits with the way I like to teach COMP-2: Using Moodle fits with my teaching style COMP-3: Moodle is compatible with most aspects of my teaching

4.95

1.40

0.92*

5.17

1.26

0.89*

4.66

1.36

0.82*

CON-1:My experience with using Moodle was better than what I expected CON-2:The benefit provided by Moodle was better than what I expected CON-3:Overall, most of my expectations from using Moodle were confirmed

4.55

1.39

0.90*

4.59

1.25

0.90*

4.75

1.14

0.82*

PU-5: Overall, using Moodle is advantageous PEOU-1: My interaction with Moodle is clear and understandable PEOU-2: Interacting with Moodle does not require a lot of mental effort PEOU-3: I find Moodle to be easy to use PEOU-4: I find it easy to get Moodle to do what I want to do

276

A.K.M. Najmul Islam Table 2. (Continued) 4.37

1.50

0.87*

4.24

1.41

0.87*

4.02

1.48

0.93*

4.53

1.35

0.75*

3,96

1.79

0.83*

4.81 4.35 4.47 4.42

1.39 1.45 1.08 1.52

0.80* 0.77* 0.74* 0.90*

PBC-1: I am able to use Moodle well for accomplishing my teaching PBC-2: Using Moodle is entirely within my control PBC-3: I have the resources necessary to use Moodle PBC-4: I have the knowledge necessary to use Moodle

5.03

1.42

0.84*

4.52

1.64

0.84*

5.24

1.51

0.85*

5.22

1.44

0.80*

SI-1: People important to me support my use of Moodle SI-2: People whose opinions I value prefer that I use Moodle for doing my teaching SI-3: In general, the school has supported the use of Moodle CR = Composite reliability; *p < 0.001

4.78

1.48

0.82*

4.53

1.37

0.85*

4.70

1.40

0.68*

Perceived Competence (CR = 0.92 )

Access (CR = 0.87)

Reliability (CR = 0.81) Perceived behavioral control (CR = 0.90)

Subjective Norm (CR = 0.83)

PC-1: I feel very competent when I use Moodle in my teaching PC-2: Most days I feel a sense of accomplishment from working with Moodle PC-3: When I am using Moodle, I often feel very capable ACCESS-1:Moodle is responsive to my request ACCESS-2:Moodle quickly loads all the text and graphics ACCESS-3:Moodle provides good access ACCESS-4:Moodle is stable REL-1:Moodle is dependable REL-2:Moodle operates reliably

Testing for discriminant validity involves checking whether the items measure the construct in question or other (related) constructs. Discriminant validity was verified with correlation analysis as recommended by Gefen and Straub [24]. The inspection of discriminant validity among variables is based on the correlation between variables and the square root of their respective average variance extracted [22]. As Table 3 shows, the squared root of the average variance extracted value for the variables is consistently greater than the off-diagonal correlation values, suggesting satisfactory discriminant validity among variables.

Antecedents of Post-adoption Expectations: A Study in the e-Learning Context

277

Table 3. Correlation among variables and squared root of average variance extracted (1) (2) (3) (4) (5) (6) (7) (8) (9) ACCESS (1) 0.79 0.58 0.88 COMP (2) 0.68 0.63 0.87 CON (3) 0.67 0.68 0.56 0.87 EOU (4) 0.44 0.65 0.44 0.69 0.84 PBC (5) 0.67 0.78 0.73 0.60 0.60 0.83 PU (6) 0.62 0.65 0.55 0.74 0.65 0.64 0.89 PC (7) 0.73 0.41 0.53 0.54 0.34 0.48 0.42 0.82 REL (8) 0.38 0.45 0.32 0.38 0.45 0.44 0.48 0.30 0.79 SN (9) Note: Compatibility (COMP), Confirmation (CON), Ease of use (EOU), Perceived behavioral control (PBC), Perceived usefulness (PU), Perceived Competency (PC), Reliability (REL), and Subjective norm (SN).

5

Results and Discussions

The test of the PLS regression model includes estimates of the path coefficients, which indicates the strengths of the relationships between the dependent and independent variables, and the R-square values, which represent the amount of variance explained by the independent variables. Fig. 3 shows the results of the test of the hypothesized model. Three out of our four hypotheses are supported. As expected, confirmation of initial expectations (β = 0.38; p < 0.001), control beliefs (β = 0.21; p < 0.05), and system characteristics (β = 0.30; p < 0.01) have a significant impact on perceived usefulness. On the other hand, and social aspect (β = 0.07, ns) have no significant effect on perceived usefulness. Taken as a whole these predictors explained 68% of the total variance in perceived usefulness. The supported hypotheses are discussed below. First, as expected confirmation affects perceived usefulness significantly. This finding is in line with the previous findings [1]. Second, we found that perceived system quality strongly predicts perceived usefulness. This finding is in line with the IS adoption [2] and success studies [13]. Our findings suggest that all the system quality related factors: perceived compatibility, ease of use, reliability, and access are important in shaping perceived usefulness. While the factors: ease of use, reliability, and access are often regarded as the system quality related attributes and found to have a direct impact on perceived usefulness, the interesting finding of our study is the importance of perceived compatibility as a system quality related variable affecting perceived usefulness. Our finding implies that educators bear distinct knowledge about their teaching situations, which they use as a basis for determining what tasks can be performed with a given elearning system. If the system is compatible with what the educators do in their teaching, the educators find the system useful. Conversely, if the system provides lot of important functionalities with a user-friendly interface but does not provide the functionalities that are needed for completion of educators’ teaching tasks, educators

278

A.K.M. Najmul Islam

Confirmation

Social norm H1:0.38*** H4: 0.07 ns Perceived usefulness H3:0.30** H2: 0.21* System quality

Control Beliefs

0.59*** PBC

0.51*** PC

0.40*** 0.322*** EOU

0.14***

Access

REL

0.32*** COMP

*p < 0.05; **p < 0.01; ***p < 0.001; ns: non-significant Fig. 3. Table 4. Fig. 3. PLS model results

would feel the system as not useful. Third, perceived control beliefs were also found to have a significant impact on perceived usefulness. Our study results revealed that both control beliefs: perceived behavioral control and perceived competency are significant to predict perceived usefulness. Our finding about the relationship between perceived competency and perceived usefulness implies that educators who perceive themselves competent enough in using the functionalities of a system are likely to feel the system as useful. Perceived competence is not so common in IS research, especially in IS adoption and use research. However, recently a number of studies are found that used perceived competence to extend expectation-confirmation based IS continuance model [6; 8]. These studies found a significant positive relationship between perceived competence and perceived usefulness. Thus, our finding is in line with the prior findings. Our finding about the relationship between perceived behavioral control and perceived usefulness implies that educators who perceive ease in using the system is likely to feel the system as useful. Our study results revealed that social norm do not have a significant impact on perceived usefulness. This is in line with the findings of Hartwick and Barki [25] where they found that although subjective norm had a significant effect on intentions prior to system development, the effect became non-significant three months after system implementation. They interpreted that before a system is developed, users’

Antecedents of Post-adoption Expectations: A Study in the e-Learning Context

279

knowledge and beliefs about a system are vague and ill-formed, and they must therefore rely more on the opinions of others as a basis for their intentions. After implementation, when more about the system’s strengths and weaknesses are known through direct experience, the subjective norm’s influence weakens. The educators in our study had prior experience with the target technology. Thus, their knowledge and beliefs about the system are concrete and eventually they do not evaluate others’ opinion to develop their post-adoption expectations.

6

Implications

Our study findings have theoretical implications. First, our study adds richness to our understanding of users’ post-adoption expectations. Second, our study revealed a number of general determinant factors that shape post-adoption expectations. We believe that there might be more general and context-specific factors that might influence post-adoption expectations. Thus, researchers should investigate more to underpin the most important factors that shape post-adoption expectations. Our study findings have a number of practical implications too. First, the strong effect of perceived system quality on post-adoption expectations provides e-learning developers and managers in planning appropriate strategies to influence users’ postadoption expectations. Our results imply that the management and developers should develop such a system that is compatible to educators’ existing values and needs. The designers should understand the educators’ teaching tasks and develop such system that is compatible to educators’ teaching tasks. In addition, they should develop such systems that are easy to use, and reliable. The service providers who host the elearning services in the Internet also should put enough effort to ensure accessibility. Second, our study also revealed that confirmation has a significant strong impact on post-adoption expectations. Thus, the management should develop strategies so that the educators may build appropriate level of initial expectations. It would allow the users to positively confirm their initial expectations. Latter, it will affect the postadoption expectations. Third, we found control beliefs have also a significant impact on post-adoption expectations. This finding implies that the management should plan for actions that might improve educators’ control beliefs. Specially, training could be arranged for the users and instruction manuals should be available to the users.

7

Conclusions

In this study, we developed a research model based on expectation-confirmation framework to underpin the factors that are salient to shape users’ post-adoption expectations. In particular, we hypothesized that confirmation, perceived control beliefs, perceived system quality, and social norm would be salient for shaping postadoption expectations. Then, we collected survey data from 175 university educators who use an e-leaning platform for teaching purpose, and analyzed the data using structured equation modeling. Study’s findings revealed that post-adoption expectation is mainly determined by confirmation of initial expectations, perceived

280

A.K.M. Najmul Islam

control beliefs, and perceived system quality. Social norm was not found to have significant impact on user satisfaction. In sum, our study revealed three main antecedents of post-adoption expectations. However, as the study was conducted in the university settings, caution should be taken before generalizing the results of this study to other context. Thus, similar kinds of studies are needed to conduct in different contexts.

References 1. Bhattacherjee, A.: Understanding Information Systems Continuance: An ExpectationConfirmation Model. MIS Quarterly 25(3), 251–370 (2001) 2. Davis, F.D.: Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly 13(3), 319–340 (1989) 3. Venkatesh, V., Bala, H.: Technology acceptance model 3 and a research agenda on interventions. Decision Sciences 39(2), 273–314 (2008) 4. Hong, S.-J., Thong, J.Y.L., Tam, K.Y.: Understanding continued information technology usage behavior: A comparison of three models in the context of mobile internet. Decision Support Systems 42(3), 1819–1834 (2006) 5. Churchill Jr., G.A., Surprenant, C.: An investigation into the determinants of customer satisfaction. Journal of Marketing Research 19(4), 491–504 (1982) 6. Roca, J.C., Gagne, M.: Understanding e-learning continuance intention in the workplace: A self-determination theory perspective. Computers in Human Behavior 24(4), 1585–1604 (2008) 7. Larsen, T.J., Sorebo, A.M., Sorebo, O.: The role of task-technology fit as users’ motivation to continue information system use. Computers in Human Behavior 25, 778–784 (2009) 8. Sorebo, O., Halvari, H., Gulli, V.F., Kristiansen, R.: The role of self-determination theory in explaining teachers’ motivation to continue to use e-learning technology. Computers & Education 53, 1177–1187 (2009) 9. Liao, C., Chen, J.L., Yen, D.C.: Theory of planning behavior (TPB) and customer satisfaction in the continued use of e-services: An integrated model. Computers in Human Behavior 23(6), 2804–2822 (2007) 10. Ajzen, I.: The theory of planned behavior. Organizational Behavior and Human Decision Processes 50(2), 179–211 (1991) 11. White, R.W.: Motivation reconsidered: The concept of competence. Psychological Review 66, 297–333 (1959) 12. Deci, E.L., Ryan, R.M.: Human autonomy: The basis for true self-esteem. In: Kemis, M. (ed.) Efficacy, Agency, and Self-Esteem, pp. 31–49. Plenum, New York (1995) 13. Seddon, P.B.: A Respecification and Extension of the DeLone and McLean Model of IS Success. Information Systems Research 8(3), 240–253 (1997) 14. Lee, J.-K., Lee, W.-K.: The relationship of e-Learner’s self-regulatory efficacy and perception of e-Learning environment quality. Computers in Human Behavior 24, 32–47 (2008) 15. Lee, S., Shin, B., Lee, H.G.: Understanding post-adoption usage of Mobile Data Services: The role of supplier-side variables. Journal of the Associations for Information Systems 10(12), 860–888 (2009) 16. Moore, G.C., Benbasat, I.: Development of an instrument to measure the perception of adopting an Information Technology innovation. Information Systems Research 2(3), 192– 222 (1991)

Antecedents of Post-adoption Expectations: A Study in the e-Learning Context

281

17. Wixom, B.H., Todd, P.A.: A Theoretical Integration of User Satisfaction and Technology Acceptance. Information Systems Research 16(1), 85–102 (2005) 18. Fishbein, M., Ajzen, I.: Belief, Attitude, Intention, and Behavior: An introduction to the Theory and Research. Addison-Wesley, Reading (1975) 19. Venkatesh, V., Davis, F.D.: A theoretical extension of the technology acceptance model: Four longitudinal field studies. Management Science 46(2), 186–204 (2000) 20. Venkatesh, V., Morris, M.G., Davis, G.B., Davis, F.D.: User acceptance of information technology: toward a unified view. MIS Quarterly 27(3), 425–478 (2003) 21. Ringle, C.M., Wende, S., Will, A.: Smart PLS 2.0 M3. University of Hamburg (2005), http://www.smartpls.de 22. Fornell, C., Larcker, D.F.: Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research 18(1), 39–50 (1981) 23. Nunnally, J.C.: Psychometric theory, 2nd edn. McGraw Hill, New York (1978) 24. Gefen, D., Straub, D.: A practical guide to factorial validity using PLS-Graph: Tutorial and annotated example. The Communications of the Associations for Information Systems 16(5), 91–109 (2005) 25. Hartwick, J., Barki, H.: Explaining the role of user participation in information system use. Management Science 40, 440–465 (1994)

Towards Reuse-Oriented and Web-Based Collaborative Framework for e-Business Providers Vladimirs Kotovs, Andrejs Lesovskis, and Leonids Novickis Division of Applied System Software, Riga Technical University, Riga, Latvia {vladimirs.kotovs,andrejs.lesovskis,leonids.novickis}@rtu.lv

Abstract. The need in systematic approach and effective software engineering methods that facilitate collaboration and allow reusing experience to address recurring problems successfully, is obvious and extremely important for e-business. The synergy of web-based and reuse-oriented methods for software solutions may create a new platform of e-business opportunities. This study describes the foundation of the framework for organizations that are moving towards a systematic reuse program and web-based information technology solutions paying special attention to the needs and the problems of e-business providers. Keywords: e-business, software reuse, web-based collaboration.

1

Introduction

Today e-business has emerged as a mainstream way to facilitate business transactions around the world. Web and Internet electronic business technologies promise to transform every aspect of the way companies conduct business and compete [1]. The Internet has fundamentally changed the economics of business transactions and provided new and different ways of conducting business. However, e-business segment of the particular domain a company is operating in is usually characterised by a very high competition. Large enterprises with their international networks have significant advantages over small and medium-sized companies. For individual small and medium enterprises (SME) it is difficult to solve the problems alone; however, by co-operation and efficient utilization of knowledge and resources, they can succeed and keep relatively big segment of the market. The technologies enable people to communicate in new ways, provide new business models, permit businesses to operate more efficiently and take advantage of the new global network economy. Companies have come to realize that if they are going to respond rapidly and effectively to todays volatile marketplace, they need to do more than web-enable discrete systems, processes or business units. They need to pull together existing systems and integrate them securely with their core business activities [1]. At the same time in the current situation of financial crisis it is extremely important for applied technologies to provide some economic and organizational T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 282–291, 2011. c IFIP International Federation for Information Processing 2011 

Towards Reuse-Oriented and Web-Based Collaborative Framework

283

benefits by reducing development and operating costs, and by efficient utilization of development knowledge and corporate expertise. The need in systematic approach and effective software engineering methods, which allow reusing experience to address recurring problems successfully, is obvious and important for e-business providers. With more than 40-year history reuse is recognized as an important mechanism to improve software quality and development productivity [2]. Today, we are just at the start of this challenge but we believe that the synergy of web-based and reuse-oriented methods for software solutions may create a new platform of e-business opportunities. The paper presents the review of the ongoing joint research activities related to the introduction of systematic software reuse and web-based collaboration in the enterprise paying special attention to the needs and the problems of e-business providers. The following sections discuss current limitations and expected benefits of the research. It is followed by the review of proposed reuse-oriented and web-based collaborative frameworks. Finally, the conclusions are made and future work is discussed in the last section.

2

Expected Benefits and the Main Drivers of the Research

The following bottlenecks and limitations have been identified as the main drivers for our joint research activities: • The existing IT solutions usually support separate stages of business processes and are not focused on integrated web-based solutions that provide an opportunity to manage core business activities within the whole company or across the entire value chain from customers to suppliers. • Intermodal collaboration among e-business service providers usually isn’t effectively organized. • Existing efforts to attain the main objective of software reuse in the ebusiness domain rendered some successful stories, however, to benefit in more systematic and repeatable way, additional research is needed. • Reuse requires up-front investments in infrastructure, methodology, training and tools without pay-offs being realized immediately and cannot be widely achieved in organizations without support of top-level management. However, reuse will only succeed, if it makes good business sense, and it will only be chosen, if a good case can be made that it is the best alternative choice for the use of capital [3]. • Reusing of large-scale system components is a problem too hard to be solved in general way. It is more effective when systematically applied in the context of families of related systems and thus is domain-dependent.

284

V. Kotovs, A. Lesovskis, and L. Novickis

The expected advances and benefits of the research are the following: • The reuse-oriented framework for organizations that are moving towards an effective reuse program composed of different aspects of software reuse and related practices. The final goal is to ingrain reuse into an organizations entire software production process [4]. A formalized process increases the chance that the project success can be repeated, facilitates adherence to the established best practices, standardization of practices across multiple projects, and helps less-expert developers to succeed via reliance on a standard process. Additionally, the framework expected to define effective measuring of economic benefits of the software reuse and development costs, where operational and strategic benefits are defined and quantified within the context of broader business strategy. However, each organization should analyse its own needs, define the key benefits, identify and remove impediments, and manage risks [5]. • The methodology for Internet-based Collaborative Framework (ICF) development is based on advanced web technologies. ICF provides the adaptation mechanism to distinct between different target groups (technology supplier, technology receiver, technology transfer facilitator). ICF is also expected to provide easy commercialization and adaptation of technology transfer process (Fig. 1).

Fig. 1. General scheme of technology transfer process

3

Reuse-Oriented Framework for e-Business Providers

Software reuse is the process of creating software systems from existing software rather than building them from scratch, whereby an organization defines a set of systematic operating procedures to specify, produce, classify, retrieve, and adapt software artefacts with the purpose of using them in its development activities [6], so that similarities in requirements and/or architecture between applications can be exploited to achieve substantial benefits [7]. Our retrospective analysis of the origins and main contributions in the research area performed in [2] outlines that software reuse is multidisciplinary and has deep and complex interactions

Towards Reuse-Oriented and Web-Based Collaborative Framework

285

Table 1. Main benefits of reuse Benefits of reuse Gains in quality Gains in reliability

Gains in productivity Gains in performance

Reduction of maintenance costs Reduction of product time to market Rapid prototyping

Description Quality of particular component could be improved because of accumulated error corrections The use of a component in several systems increases the chance of errors to be detected and improves confidence in that component Productivity could be achieved due to less design and code to be developed Extensive reuse can be worth the effort invested in optimizations, that may yield better performance of a reused component Fewer defects and reduction of maintenance costs can be expected when proven quality components have been used By using reusable artifacts organizations could reduce the product time to market which influences the success or failure of a software product Reusable components can provide an effective basis for quick building of a prototype of a software system

with many areas of computer science. Table 1 consolidates the main benefits from [8] that can be achieved by the organizations with the maximization of the reuse level. Directions and components for reuse-oriented framework comply with the following main conclusions that we have obtained through the survey [9] involving software organizations in Baltic regions, whose responses were analysed and used to relate the characteristics of organizations with their reuse experience: • 72% participants claimed to succeed in projects by the means of software reuse in their organization, and small and medium software organizations presented higher reuse success rates than organizations of a large size. • Organizations should focus on the development of product families if applicable in the operating business area. • Attention should be spent on the introduction of process-driven systematic reuse in the organization. • Organizations should consider using repository for storing and retrieving reusable assets, and a configuration management process should guarantee proper evolution of these assets. The overall structure of the considered reuse-oriented framework is organized in economic (A), organizational (B), and process (C) dimensions as outlined on Figure 2.

286

V. Kotovs, A. Lesovskis, and L. Novickis

Fig. 2. The overall structure of reuse-oriented framework

(A) From the economic perspective implementing a reuse initiative in a corporate environment requires a decision about when and where capital investment for software reuse is to be made, and whether it will be proactive or reactive [3]. The measuring of economic benefits of the software reuse and development costs should be aligned with the following main principles from [10]: • Economic value maximization drives reuse investment strategies for the business. • Strategy drives selection of reuse investments that are actively structured to maximize embedded strategic options. Economic aspects of software reuse are divided into the following broad classes by [6]: – Metrics, which reflect attributes that increase the market value of an asset – Reuse cost estimation techniques and return-on-investment (ROI) models to quantify reuse related decisions. (B) Several approaches exist for establishing a reuse program from the organizational perspective: • Centralized, with an organizational unit dedicated to developing, distributing, maintaining, and providing training on reusable assets. • Distributed asset development, where reuse is implemented collaboratively by projects in the same product line. Reuse environment on the organizational level assumes development of – Managerial infrastructure in the form of a set of functions, responsibilities, reporting requirements, and reward, which are required to ensure operation of reuse processes – Technological infrastructure that includes a configuration management and quality assurance functions to support reuse operations, as well as enforcement of testing, verification and asset certification standards.

Towards Reuse-Oriented and Web-Based Collaborative Framework

287

(C) Process-driven means that the software development is done in accordance with well defined processes that are enforced through management policies. A software process could be defined as a set of activities that lead to the production of a software product [11], and it is important in order to ensure efficiency, reproducibility, homogeneity, and predictable time and effort constraints. As it was noted already, a key concept of systematic reuse is the domain, which may be defined as an application area or, more formally, a set of systems that possess similar functionality and share design decisions. By domain-specific reuse we assume, that the reusable assets, the development processes, and the supporting technologies are appropriate to the application domain for which the software is being developed. It is common to identify three main stakeholders in the process of software reuse within an organization: corporate management, domain engineering and application engineering teams. By reviewing the tasks of each of these stakeholders, the following top-level reuse related activities can be identified: • Asset production - identification, development and classification of reusable artefacts • Asset usage - locating and evaluating assets, and achieving their actual reuse by integrating them into applications being developed • Asset management - asset storage, repository management, and asset dissemination • Maintenance and support - support for asset usage, methodological support, and corrective and evolutionary maintenance of assets • Reuse management - introducing and monitoring reuse within a company or department. To support reuse, software processes have to consider two facets: developing for reuse and developing with reuse. As a result comparing to conventional monolithic set of tasks organizations are responsible for providing and maintaining software systems, software reuse introduces a differentiation between the tasks related to the production of reusable assets and the tasks related to the production of end-user applications [12]. Such two-life-cycle approach with generalized activities schematically displayed on Figure 3 is commonly referred as domain and application engineering, which address development for reuse and development with reuse respectively. From the technical point of view the e-business provider development within the proposed framework should rely on the best practices and appropriate supporting methodologies facilitating global visibility, integration and collaboration as noted in section 1. The Internet-based collaborative framework (ICF) as discussed in the next section aimed to provide a facade for several advanced IT solutions in order to support e-business.

288

V. Kotovs, A. Lesovskis, and L. Novickis

Fig. 3. The two-life-cycle model of domain and application engineering [13]

4

Internet-Based Collaborative Framework for e-Business Providers

E-business denotes using computer-mediated networks to perform a wide range of commercial transactions, to automate and transform business processes at the firm level, and to manage more effectively value chains across firms. In addition to e-commerce, customer service and support applications, e-business can include human-resource functions, sales-force automation, supply-chain management and other applications to automate firms’ central internal processes [14]. The online environment that supports dynamic e-business should provide the following capabilities: • Open application interfaces that provide interaction between business partners, service providers, and other participants; • Open standards that define business transaction processes and would improve automation of business to business (B2B) interactions; • Internet-based communication methods with opportunity to extend to other public or private systems; • Easy adaptation to market condition changes. Web services enable many of these key prerequisites for dynamic e-business solutions. They define a standard way to describe and invoke a Web service interface so that the business application can easily connect to it. In the report entitled Executive Survey: SOA Implementation Satisfaction, Hurwitz & Associates analysts point out that the top reasons for SOA adoption include the expectation of greater reuse in existing and newly-built Web services, business flexibility, ease

Towards Reuse-Oriented and Web-Based Collaborative Framework

289

of integration and speed of integration with nearly 90% of respondents pointing to service reuse as their number one concern [16]. According to World Wide Web Consortium (W3C), there are two major classes of Web services: • REST-compliant Web services, in which the primary purpose of the service is to manipulate XML representations of Web resources using a uniform set of ”stateless” operations; and • Arbitrary Web services, in which the service may expose an arbitrary set of operations (most of the SOAP-based Web services) [15]. REST (Representation State Transfer) provides four familiar operations GET, POST, PUT, and DELETE of the HTTP standard, and make the resource addressing handle the variability. SOAP (Simple Object Access Protocol), on the other hand, permits quite a complex set of XML formatted commands and data transmission options. In contrast with REST which requires HTTP, SOAP messages can be moved by any transport method which can handle Unicode text. But HTTP transport is preferred by many developers because it is more familiar and SOAP with HTTP can tunnel easily over existing firewalls and proxies, without modifications to the SOAP protocol, and can use the existing infrastructure. Technologies like SOA, Web services, Semantic Interoperable Services, XML, SOAP, and AJAX Web development techniques can be used as a technological platform for the integration. Using these technologies together with the existing enterprise web platforms and portal frameworks (like Liferay, GateIn, WebSphere Portal, JetSpeed, etc), it is possible to design the integrated portals that will provide seamless integration of the various business-related services and data sources. For instance, separate portals that support different logistics business processes along the selected freight route (www.elogmar-m.org, ww.containeronlineshop.de, www.e-56.com.cn) are united into one framework with a single entry point (Fig. 4) [17]. An enterprise portal, built with a robust portal product, provides an off-theshelf framework for developing and deploying service-oriented applications. It can serve as a ready-made Web services consumer platform, and enables you to build composite applications, deploy syndicated content from other portals through remote portlets, replace/augment legacy interfacing applications, create common views of data, and facilitate access by mobile and wireless devices [18]. Most of the mentioned above portal frameworks also have personalization features. Personalization provides end-users with an opportunity to customize their version of a portal; thereby, greatly improving end-user performance and satisfaction. Using personalization mechanism, portal administrators can define specific groups which may represent portal users real-life roles (operator, manager, CEO, etc). Depending on their roles, users will have access to the features and content designed for their group. One of the goals in our test case is the integration of eLogmar portal services with a newly designed portal. We are using NuSOAP Toolkit in eLogmar portal

290

V. Kotovs, A. Lesovskis, and L. Novickis

Fig. 4. Integration in ICF

to provide SOAP-based access to system functionality (shipping lines services, freight cost calculation, etc).

5

Conclusion

The need in systematic approach and effective software engineering methods that facilitate global visibility and collaboration and allow reusing experience to address recurring problems successfully, is obvious and extremely important for e-business. The study presents the foundation of the framework for organizations moving towards a reuse program that is organized considering economic, organizational and process aspects of software reuse. Although the work might be still considered to be in its early stages, the study defines the basis, the components and directions for further research activities. Based on our experience in portal development and integration projects, we are currently working on consolidation of best practices and general recomendations for e-business providers. Another important issue concerns maturity of reuse processes without gaps in reuse activities covering steps of domain engineering, domain analysis, design and implementation. The planned future work is related to the formalization of the processual dimension of the proposed framework and methodolodical support of ICF. More attention is planned to be paid to the processes of identification of domain parts reasonable for automation and to systematic software reuse opportunities in the context of Model-Driven Development. Acknowledgments. This work is partly financed by the European Union (European Regional Development Fund) within the Baltic Sea Region Programme 2007-2013 project BONITA (Baltic Organisation and Network of Innovation Transfer Associations). The present activity is partly funded by the ERDF

Towards Reuse-Oriented and Web-Based Collaborative Framework

291

project ”Support of FP7 ICT STREP project ”Simulation Highway””. This work has been supported by the European Social Fund within the project ”Support for the implementation of doctoral studies at Riga Technical University”.

References 1. Westermeier, J., Plave, L.J.: E-Business: The E-Business Legal Survival Kit (2004) 2. Kotov, V.: Forty years of software reuse. Scientific Journal of RTU 38, 153–160 (2009) 3. Frakes, W.B., Kang, K.C.: Software Reuse Research: Status and Future. IEEE Transactions on Software Engineering 31(07), 529–536 (2005) 4. Frakes, W.B.: Software Reuse and Domain Engineering, http://wfrakes.wordpress.com/category/introduction 5. Frakes, W.B., Isoda, S.: Success Factors of Systematic Reuse. IEEE Softw. 11, 14–19 (1994) 6. Mili, H., Mili, A., Yacoub, S., Addy, E.: Reuse-based Software Engineering: Techniques, Organization, and Measurement. John Wiley & Sons Inc. (2002) 7. Ezran, M., Morisio, M., Tully, C.: Practical Software Reuse. Springer, Heidelberg (2002) 8. Sametinger, J.: Software Engineering with Reusable Components. Springer, Heidelberg (1997) 9. Kotov, V.: Reuse In Software Development Organizations In Latvia. Scientific Journal of RTU. 5 43, 90–96 (2010) 10. Favaro, J., Favaro, K., Favaro, P.: Value Based Software Reuse Investment. Annals of Software Engineering 5, 5–52 (1998) 11. Sommerville, I.: Software engineering, 6th edn. Addison-Wesley (2001) 12. Ezran, M., Morisio, M., Tully, C.: Practical Software Reuse. Springer, Heidelberg (2002) 13. Linden, F., Schmid, K., Rommes, E.: Software Product Lines in Action: The Best Industrial Practice in Product Line Engineering. Springer, Heidelberg (2007) 14. Canadian e-Business Initiative, The Digital Economy in Canada: The e-Business Environment (2004), http://www.ic.gc.ca/eic/site/ecic-ceac.nsf/eng/gv00501.html 15. W3C Working Group, Web Services Architecture (2004), http://www.w3.org/TR/ws-arch/ 16. Baroudi, C., Halper, F.: Executive Survey: SOA Implementation Satisfaction (2006), http://www.cio.co.uk/cmsdata/whitepapers/4474/110206_Mindreef_Report 17. Bluemel, E., Strassburger, S., Novitsky, L.: IT&T Solutions in Logistics and Maritime Applications. JUMI (2006) 18. Phifer, G.: A Portal May Be Your First Step to Leverage SOA, http://www.gartner.com/DisplayDocument?doc_cd=13014

Weibull Model for Dynamic Pricing in e-Business Nicholas Nechval1, Maris Purgailis1, and Konstantin Nechval2 1

University of Latvia, EVF Research Institute, Statistics Department, Raina Blvd 19, LV-1050 Riga, Latvia [email protected] 2 Transport and Telecommunication Institute, Applied Mathematics Department, Lomonosov Street 1, LV-1019 Riga, Latvia [email protected]

Abstract. As is the case with traditional markets, the sellers on the Internet do not usually know the demand functions of their customers. However, in such a digital environment, a seller can experiment different prices in order to maximize his profits. In this paper, we develop a dynamic pricing model to solve the pricing problem of a Web-store, where seller sets a fixed price and buyer either accepts or doesn’t buy. Frequent price changes occur due to current market conditions. The model is based on the two-parameter Weibull distribution (indexed by scale and shape parameters), which is used as the underlying distribution of a random variable X representing the amount of revenue received in the specified time period, say, day. In determining (via testing the expected value of X) whether or not the new product selling price c is accepted, one wants the most effective sample size n of observations X1, …, Xn of the random variable X and the test plan for the specified seller risk of Type I (probability of rejecting c which is adequate for the real business situation) and seller risk of Type II (probability of accepting c which is not adequate for the real business situation). Let μ1 be the expected value of X in order to accept c, and μ2 be the expected value of X in order to reject c, where μ1>μ2, then the test plan has to satisfy the following constraints: (i) Pr{statistically reject c | E{X} = μ1} = α1 (seller risk of Type I), and (ii) Pr{statistically accept c | E{X} = μ2} = α2 (seller risk of Type II). It is assumed that α1 < 0.5 and α2 < 0.5. The cases of product pricing are considered when the shape parameter of the two-parameter Weibull distribution is assumed to be a priori known as well as when it is unknown. Keywords: e-business, pricing, uncertainty, revenue, Weibull model, seller risk, test plan.

1

Introduction

The benefits of dynamic pricing are twofold. First, it provides new opportunities for companies to maximize their return per customer. With lower menu costs (that is, the cost of displaying prices to customers), companies can have multiple prices for different channels and product configurations − and can change those prices more T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 292–304, 2011. © IFIP International Federation for Information Processing 2011

Weibull Model for Dynamic Pricing in e-Business

293

frequently. Companies able to gather information about their competition and about customer needs and willingness to pay can customize their offerings and prices. This enables them to deploy dynamic pricing through the most appropriate of many channels. With dynamic pricing, companies can give their customers exactly what they want, at exactly the price they are willing to bear. Nothing is left on the table. The second, perhaps less obvious, benefit is that dynamic pricing can also bring better returns on deployed assets. For businesses with high fixed-cost technology infrastructures, periods of low demand and, thus, low utilization are expensive. Conversely, when there are inflexibilities in the supply chain for critical components, periods of high demand can lead to shortages and can both delay purchasing and damage customer relationships. But with dynamic pricing, companies can encourage demand in slow periods and discourage it in busy periods. Pricing in e-business is similar to that in traditional business. However, in ebusiness prices are readily available to consumers and competing organizations via the Internet. With vital pricing data readily available, organizations may be forced to reevaluate their pricing methodology and strategy to avoid being significantly undercut, while remaining competitive in their respective markets. On Internet markets, as is the case on most traditional markets, sellers do not usually know the demand functions of their customers without ambiguity. So as to establish a reasonable theory of the way Web-stores should fix their prices on the network, it is thus necessary to take the uncertainty they face into account by explicitly introducing the process by which they acquire information on their environment. In most uncertain environments, Web-stores can only learn what the most profitable prices are by price experimentations. By its very nature, the Internet is well adapted to such a learning process. Indeed, in this electronic environment the menu costs are negligible and sellers can therefore easily experiment different prices. However, as such a learning activity is costly, the Web-stores have to make a tradeoff between charging the most profitable price according to their current information (i.e., to exploit their information) and inquiring on the profitability of the other prices (i.e., to explore the profitability of the other prices). The determination of the Web-stores’ pricing strategy under uncertainty is thus an inter-temporal decision problem in which they have to make an exploration vs. exploitation compromise. Some dynamic pricing algorithms have already been proposed in the E-commerce literature. Much recent work in dynamic pricing has concentrated on pricing perishable products, such as are typically sold by the airline, hotel, car rental, and fashion industries. Inventory levels are a critical consideration in pricing such products. A large literature has addressed dynamic pricing in the presence of inventory considerations, where it is usually referred to as “revenue management” or “yield management.” A recent comprehensive review of such practices is given in [1]; also see [2]. In a classic paper, Rothschild [3] investigated the nature of optimal sequential pricing strategies that experiment with two prices. He showed that under a general class of prior distributions on demand, there is a positive probability that a pricing policy that maximizes expected total discounted reward over an infinite horizon will select the revenue-maximizing price only finitely many times. Therefore, the policy may converge to the suboptimal price. This phenomenon was studied more generally by Easley and Kiefer [4, 5], as well as McLennan [6] and Aghion et al. [7]. Several

294

N. Nechval, M. Purgailis, and K. Nechval

other prior studies have also described the qualitative behavior of optimal priceadjustment strategies using parametric classes of demand functions, such as those by Grossman, Kihlstrom, and Mirman [8], Mirman, Samuelson, and Urbano [9]. Leloup and Deveaux [10] consider pricing strategies in a Bayesian framework. In this paper, however, we shall consider markets where inventories are not as important to pricing and focus on the revenue implications of demand functions online. By taking a frequentist approach, our methods contrast with some line search-type methods previously proposed for this problem in the literature, such as the “derivative following” methods proposed by Greenwald and Kephart [11], Dasgupta and Das [12]. These methods resemble stochastic approximation algorithms for locating the revenuemaximizing price. While simple to implement, these methods do not incorporate prior information, nor do they make full use of the data that have been observed.

2

Model Formulation

Consider a seller (vendor) operating a shop where a single product is for sale. Here, a shop is understood to be any space, either real or virtual, where prices are displayed to arriving customers, who then decide whether to purchase the product. For example, a shop may be a web page or collection of web pages that display the price for a particular product, such as a software product or digital media document, and provides visitors with links to purchase the product. We assume that information about present and past prices is not available outside of the shop, as would be good practice if the seller wishes to strictly control the price of the product. In each of consecutive time periods of equal length (say, day), a random number of customers arrive to the shop, observe the posted price, and individually decide to either purchase some quantity of the product or exit the shop without purchasing. We shall further assume that the numbers of customers arriving in each period form an identically distributed sequence that is independent of past arrivals as well as both past and current prices. This assumption may be warranted in a finite time horizon if we assume that the customer base is composed, for example, of a large number of distributed Internet users, each of whom desires the good in a given period with a small probability, independent of other users and independent of past demand. We also assume that the number of consecutive time periods of equal length represents a deterministic time horizon (which may be infinite if no definite end period is anticipated). The seller may change the price at the start of any time period, always choosing from a predetermined set of prices. Given that the price is only displayed in the shop, we assume there is no cost to changing the price. A customer will buy the product if the current price is less than or equal to his or her privately held reservation price. We shall assume that the reservation prices of arriving customers also form an identically distributed sequence that is independent of past prices, the reservation prices of past customers, and past numbers of customer arrivals. In some time period, which consists of n consecutive time periods of equal length, the seller sets a new price c for product and receives a random amount of revenue according to the number of customers arriving to the shop in that period whose

Weibull Model for Dynamic Pricing in e-Business

295

reservation prices are at least as great as c. The seller can only observe a random amount of revenue, X, received in each of n consecutive time periods of equal length. Thus, he will have available a sample of observations X=(X1, ⁄, Xn). It is assumed that the random variable X follows the two-parameter Weibull distribution with the probability density function

f (x | β ,δ ) =

δ β

x  β

  

δ −1

 x exp −    β

  

δ

  ( x > 0) 

(1)

indexed by scale and shape parameters β and δ. The problem is to find the sample size n of observations of the random variable X and test plan based on X=(X1, ⁄, Xn) in order to determine whether or not the new product selling price c is accepted. The test plan has to satisfy the specified constraints for the seller risks of Type I and II, respectively.

3

Test Plan When the Shape Parameter Is a Priori Known

In determining whether or not the new product selling price c is accepted, we have to use the test plan which satisfies the following conditions: Pr{statistically reject c | E{ X } = μ1} = α1 (seller risk of Type I),

(2)

Pr{statistically accept c | E{ X } = μ 2 } = α 2 (seller risk of Type II),

(3)

where ∞

1  E{ X } = μ = xf ( x | β , δ )dx = β Γ1 +  δ   0



(4)

is the expected value of X, E{X}=μ1 is statistically acceptable for c, E{X}=μ2 is statistically rejectable for c, α1 and α2 are the specified seller risks of Types I and II, respectively. To test for assuring μ in the Weibull distribution, let n observations of the random variable X be made and the complete data sample X=(X1, X2, ⁄, Xn) be obtained. Then, if the shape parameter δ is known a priori, the maximum likelihood estimator of β δ is given by 

n

β δ =  X iδ / n,

(5)

 V = 2n ( β δ / β δ )

(6)

i =1

and the pivotal quantity

is χ 2 distributed with 2n degrees of freedom. The n and h are determined by

296

N. Nechval, M. Purgailis, and K. Nechval

 β δ > h, accept c,  δ  β ≤ h, reject c.

(7)

To select n and h, we use the equations (2) and (3), which can be presented in the case when the parameter δ is known as δ δ 2nh   2nβ Pr β ≤ h | μ = μ1 ; δ = Pr  δ ≤ δ | μ = μ1; δ  = Pr V ≤ χ 22n;α1 = α1, (8)   β1 β1 δ    2nβ 2nh Pr β δ > h | μ = μ 2 ; δ = Pr  δ > δ | μ = μ 2 ;δ  = Pr V > χ 22n;1−α 2 = α2, (9)   β 2 β2 where

{

{

}

{

}

{

βj =

μj

Γ(1 + 1 / δ )

,

j = 1, 2,

}

}

(10)

χ 22n; p is the p-quantile of χ 2 distribution with 2n degrees of freedom. From (8) and (9),

χ 22n;α1 β 2δ ≤ . β1δ χ 22n;1−α 2

(11)

Therefore, the required n is the smallest integer satisfying (11). Using n based on (11),  we have from (8) an (10) that the h for β δ is given by δ

2   χ 2 n;α1 μ1  (12) . h =  2n  Γ(1 + 1 / δ )   Thus, the new product selling price c is accepted if β δ > h; otherwise, it is rejected.

4

Test Plan When the Shape Parameter Is Unknown

4.1

Preliminaries

We consider both parameters β, δ to be unknown. We will give here appropriate expressions for the case of complete (uncensored) sampling. Let X = (X1, …, Xn) be a   random sample from the two-parameter Weibull distribution (1), and let β , δ be maximum likelihood estimates of β, δ computed on the basis of X = (X1, …, Xn). In terms of the Weibull variates, we have that     V1 = ( β / β )δ , V2 = δ / δ , V3 = ( β / β )δ (13) are pivotal quantities. Furthermore, let   Z i = ( X i / β )δ ,

i=1, …, n.

(14)

Weibull Model for Dynamic Pricing in e-Business

297

It is readily verified that any n−2 of the Zi’s, say Zi, …, Zn-2 form a set of n−2 functionally independent ancillary statistics. The appropriate conditional approach, first suggested by Fisher [13], is to consider the distributions of V1, V2, V3 conditional on the observed value of Z = (Zi, …, Zn). (For purposes of symmetry of notation we include all of Zi, …, Zn in expressions stated here; it can be shown that Zn, Zn-1, can be determined as functions of Zi, …, Zn-2 only.) The marginal distribution of  V2 = δ / δ , (15) conditional on fixed z=(zi, …, zn) is appropriate for making inference about δ when β is unknown. It can be found (see Corollary 1.1 below with r = n) that the marginal probability density function of V2, conditional on fixed z, is given by

g 2 (v2 | z ) = ϑ (z )v2n − 2

n

∏ i =1

 n ziv 2  ziv 2   i =1



−n

  , (v2 > 0),  

(16)

where −n ∞  n   n ϑ (z ) =  v2n − 2 ziv 2  ziv 2  dv2      i =1   i =1 0 



∏



−1

(17)

is the normalizing constant. The probability statement Pr{q1 ≤ V2 ≤ q2 | z} = 1 − α

(18)

leads in the usual way to the (1−α) conditional confidence interval   q1δ ≤ δ ≤ q2δ .

(19)

The marginal distribution of   V3 = ( β / β )δ

(20)

is used to make inferences about β, when δ is unknown. It can be found (see Corollary 2.1 below with r=n) that the probability distribution function of V3, conditional on fixed z, can conveniently be expressed as Pr{V3 ≤ h | z} = ϑ (z )

n−2 ∞ v2  



 ∏i =1 n

 

0

ziv 2 Gn  h v 2  



n

z v2 i =1 i

 

i =1 ziv

n

n

2

  dv

2

(h ≥ 0),

(21)

where Gn(τ) represents the incomplete gamma integral, Gn (τ ) =

τ

t n−1e −t dt. Γ ( n) 0



(22)

298

N. Nechval, M. Purgailis, and K. Nechval

For any specified value of h, (21) can be integrated numerically to give conditional probability statements for V3. Then, for example, the probability statement Pr{V3 ≤ h | z} = α leads to an α conditional confidence interval for β,   β ≥ h −1 / δ β .

(23)

(24)

Main Theorems. The results can be presented more naturally, however, if we consider the distribution of the logarithm of a Weibull variate, which we denote by Y. The random variate Y follows the first asymptotic distribution of extreme values, with density  1  y−a  y − a  f ( y | a, b) = exp (25)  exp − exp   (−∞ < y < ∞), b b    b   where a = ln β ,

b = δ −1.

(26)

Now, (25) is a distribution with location and scale parameters a and b, respectively,   and it is well known that if a , b are maximum likelihood estimates for a, b, from a complete (uncensored) sample of size n, then     U 1 = ( a − a ) / b, U 2 = b / b, U 3 = ( a − a ) / b (27) are pivotal quantities, whose distributions depend only on n. That is, inferences concerning a and b here may be based on the pivotal quantities (27) as usual; however, the distributions of these pivotals should be considered conditional on the observed values of ancillary statistics. We generalize this situation and allow the original data to be possibly censored. Here the following theorems hold. Theorem 1. Let Y1 ≤ ... ≤ Yr be the first r ordered past observations from a sample of size n from the first asymptotic distribution of extreme values (25). Then the marginal probability density function of the pivotal quantity  U 2 = b / b,

(28)

s = ( s1 , ..., sr ) ,

(29)

  Si = (Yi − a ) / b , i = 1, ..., r ,

(30)

conditional on fixed

where

 are ancillary statistics, any r−2 of which form a functionally independent set, a and  b are the maximum likelihood estimators of a and b based on the first r ordered past

Weibull Model for Dynamic Pricing in e-Business

299

observations (Y1≤ ... ≤Yr) from a sample of size n from the first asymptotic distribution of extreme values (25), which can be found from solution of       r a = b ln  e yi / b + (n − r )e y r / b  r ,      i =1



(31)

and     r    r b =  yi e yi / b + (n − r ) y r e y r / b  e yi / b + (n − r )e y r / b      i =1  i =1  is given by



g 2 (u 2 | s)



= ϑ (s)u2r − 2

−1

−

1 r

r

 yi ,

(32)

i =1

−r

r   r  exp u 2 si   exp( si u2 ) + (n − r ) exp( sr u 2 ) ,     i =1   i =1





u2 ∈ (0, ∞),

(33)

where −r ∞  z   r   r −2   ϑ (s) =  u 2 exp u2 si   exp( si u 2 ) + (n − r ) exp( sr u 2 ) du2  0    i =1   i =1   is the normalizing constant.





−1



(34)

Proof. The joint density of Y1 ≤ ... ≤ Yr is given by f (y1, ...,yr | a,b) =

r  n! 1  yi − a  y −a    y − a  − exp i   exp − (n − r ) exp r exp  . (35) (n − r )! i =1 b  b  b   b  

∏

Using the invariant embedding technique [14-21], we then find in a straightforward manner, that the probability element of the joint density of U1, U2, conditional on fixed s = ( s1 , ..., sr ) , is f (u1 , u 2 | s)du1du2 r     r  = ϑ  (s)u 2r − 2 exp u2 si e ru1 exp − e u1  exp( si u2 ) + (n − r ) exp( sr u 2 ) du1du 2 ,     i =1    i =1  





u1∈(−∞, ∞), u2∈(0, ∞),

(36)

where −1

−r ∞  z   r    r −2   ϑ (s) =  Γ(r )u2 exp u 2 si   exp( si u2 ) + (n − r ) exp( sr u 2 ) du2  (37) 0    i =1   i =1   is the normalizing constant. Now u1 can be integrated out of (36) in a straightforward  way to give (33). This ends the proof.







300

N. Nechval, M. Purgailis, and K. Nechval

Corollary 1.1. Let X1 ≤ ... ≤ Xr be the first r ordered past observations from a sample of size n from the two-parameter Weibull distribution (1). Then the marginal probability density function of the pivotal quantity

 V2 = δ / δ ,

(38)

z=(zi, …, zn),

(39)

  Z i = ( X i / β )δ , i = 1, ..., r ,

(40)

conditional on fixed

where

 are ancillary statistics, any r−2 of which form a functionally independent set, β and  δ are the maximum likelihood estimators of β and δ based on the first r ordered past observations (X1≤ ... ≤Xr) from a sample of size n from the two-parameter Weibull distribution (1), which can be found from solution of  1/ δ

    r  β =   xiδ + (n − r ) xrδ  r       i =1

,

(41)

and −1

−1  r      r  1 r δ =   xiδ ln xi + (n − r ) xrδ ln x r   xiδ + (n − r ) xrδ  −  ln xi  ,  i =1  r i =1  i =1   



(42)

is given by −r

r  r  g 2 (v2 | z ) = ϑ (z )v2r − 2 ∏ ziv 2   ziv 2 + (n − r ) z rv 2  , (v2 > 0),   i =1  i =1 

(43)

where −r ∞  r r v2  v2 v2  r −2    ϑ (z ) =  v2 ∏ zi   zi + (n − r ) z z  dv2    i =1  i =1  0 

−1

(44)

is the normalizing constant. Theorem 2. Let Y1 ≤ ... ≤ Yr be the first r ordered past observations from a sample of size n from the first asymptotic distribution of extreme values (25). Then the probability distribution function of the pivotal quantity   U 3 = (a − a ) / b ,

(45)

Weibull Model for Dynamic Pricing in e-Business

301

conditional on fixed s, is given by Pr(U 3 ≤ h | s)  r   r   r −2  u2  si Gr  exp(hu2 ) exp(si u2 ) + (n − r ) exp(sr u2 )  u exp 2 ∞      i =1    i =1  

= ϑ (s) 

  r  exp(si u2 ) + (n − r ) exp(sr u2 )   i =1

0

r

du2 (h ≥ 0),

(46) where τ

t r −1e −t dt. Γ( r ) 0

Gr (τ ) = 

(47)

represents the incomplete gamma integral. 

Proof. The proof is similar to that of Theorem 1 and so it is omitted here.

Corollary 2.1. Let X1 ≤ ... ≤ Xr be the first r ordered past observations from a sample of size n from the from the two-parameter Weibull distribution (1). Then the probability distribution function of the pivotal quantity   V3 = ( β / β )δ

(48)

conditional on fixed z, is given by

Pr{V3 ≤ h | z} = ϑ (z )

r −2  ∞ v2 



zv  ∏i =1 i

0

r

 

2

G  h v 2     r



r

z v2 i =1 i

i =1 ziv r

2

+ (n − r ) z rv 2

+ (n − r ) z rv 2     dv  

r

2

(h ≥ 0). (49)

4.2

Test Plan

The n and h are determined by

 β δ > h, accept c,   δ  β ≤ h, reject c.

(50)

To select n and h, we use the equations (2) and (3), which can be presented in the case when the parameter δ is unknown as

302

N. Nechval, M. Purgailis, and K. Nechval

  β δ  h Pr β ≤ h | μ = μ1; z = Pr   ≤  | μ = μ1 ; z  = Pr V3 ≤ V3( z ( n );α1 ) = α1, (51)  β1δ  β1δ   β δ  δ h Pr β > h | μ = μ 2 ; z = Pr   >  | μ = μ 2 ; z  = Pr V3 > V3( z ( n );1−α 2 ) = α2, (52) δ δ  β 2  β2

{

}

 δ

{

}

{

}

{

}

where Pr{V3 ≤ V3( z ( n );α ) | z} = α .

(53)

From (51) and (52), 

β 2δ



δ

β1

≤

V3( z ( n );α1 ) V3( z ( n );1−α 2 )

.

(54)

Therefore, the required n is the smallest integer satisfying (54). Using n based on (54),  we have from (51) and (10) that the h for β δ is given by 

δ

   1   h =  μ1 / E Γ1 +    V3( z ( n );α1 ) ,    δ V2   

(55)

where ∞    1  1  E Γ1 +   = Γ1 +   g 2 (v2 | z )dv2 . (56)   δ V2  0  δ v2   Thus, the new product selling price c is accepted if β δ > h; otherwise, it is rejected.



5

Conclusions and Directions for Future Research

The choice of an appropriate dynamic pricing model is a crucial decision problem because a good model not only can help sellers save testing time and reduce testing cost, but it also can positively affect the image of the product and thus attract more consumers to buy this product. Therefore often the Bayesian approach is used to solve the above problem. Unfortunately, in this case the subjectivity of investigator (a limitation of the Bayesian approach) is introduced through a priori distribution. In order to rule out the subjectivity of investigator and to consider comprehensively the relevant risks, in this paper a frequentist (non-Bayesian) decision analysis is employed. It is based on the invariant embedding technique [14-21] which allows one to use available statistical information as completely as possible. To extend the scope of our analysis, it would be interesting to explicitly introduce price competition between Web-stores. Indeed, the Internet environment may be a

Weibull Model for Dynamic Pricing in e-Business

303

very competitive market and the sellers have to take the behaviors of their competitors into account in their own dynamic pricing strategy. So as to take this competition into account in our model, we can introduce a non-stationary behavior of the customers on the market. Such an extension would be a very big step towards a generalization of the pricing model proposed in this paper.

Acknowledgments. This research was supported in part by Grant No. 06.1936, Grant No. 07.2036, Grant No. 09.1014, and Grant No. 09.1544 from the Latvian Council of Science and the National Institute of Mathematics and Informatics of Latvia. The authors are also very grateful to two anonymous referees for their detailed comments that have helped in improving an earlier version of this paper.

References 1. Elmaghraby, W., Keskinocak, P.: Dynamic Pricing in the Presence of Inventory Considerations: Research Overview, Current Practices and Future Directions. Manage. Sci. 49, 1287–1309 (2003) 2. McGill, J.I., van Ryzin, G.J.: Revenue Management: Research Overview and Prospects. Transport Sci. 33, 233–256 (1999) 3. Rothschild, M.: A Two-Armed Bandit Theory of Market Pricing. J. Econ. Theo. 9, 185– 202 (1974) 4. Easley, D., Kiefer, N.M.: Controlling a Stochastic Process with Unknown Parameters. Econometrica 56, 1045–1064 (1988) 5. Easley, D., Kiefer, N.M.: Optimal Learning with Endogenous Data. Int. Econ. Rev. 30, 963–978 (1989) 6. McLennan, A.: Price Dispersion and Incomplete Learning in the Long Run. J. Econ. Dynam. Contr. 7, 331–347 (1984) 7. Aghion, P., et al.: Optimal Learning by Experimentation. Rev. Econ. Stud. 58, 621–654 (1991) 8. Grossman, S.J., Kihlstrom, R.E., Mirman, L.J.: A Bayesian Approach to the Production of Information and Learning by Doing. Rev. Econ. Stud. 44, 533–547 (1977) 9. Mirman, L.J., Samuelson, L., Urbano, A.: Monopoly Experimentation. Int. Econ. Rev. 34, 549–563 (1993) 10. Leloup, B., Deveaux, L.: Dynamic Pricing on the Internet: Theory and Simulation. J. Electron. Comm. Res. 1, 53–64 (2001) 11. Greenwald, A.R., Kephart, J.O.: Shopbots and Pricebots. In: Proceedings of the 5th International Conference on Artificial Intelligence (IJCAI 1999), pp. 506–511. Morgan Kaufmann, San Mateo (1999) 12. Dasgupta, P., Das, R.: Dynamic Pricing with Limited Competitor Information in a Multiagent Economy. In: 4th International Conference on Coordination Languages and Models (COORDINATION 2000). Springer, New York (2000) 13. Fisher, R.A.: Two New Properties of Mathematical Likelihood. Proceedings of the Royal Society A 144, 285–307 (1934) 14. Nechval, N.A., Nechval, K.N., Vasermanis, E.K.: Effective State Estimation of Stochastic Systems. Kybernetes 32, 666–678 (2003) 15. Nechval, N.A., Vasermanis, E.K.: Improved Decisions in Statistics. SIA, Izglitibas (2004)

304

N. Nechval, M. Purgailis, and K. Nechval

16. Nechval, N.A., Nechval, K.N.: Constrained Optimization in Newsboy Problems under Uncertainty via Statistical Inference Equivalence Principle. In: Al-Begain, K., Bolch, G., Telek, M. (eds.) Proceedings of the 12th International Conference on Analytical and Stochastic Modelling Techniques and Applications (ASMTA 2005), pp. 166–171. RTU, Riga (2005) 17. Nechval, N.A., Berzins, G., Purgailis, M., Nechval, K.N.: Improved Estimation of State of Stochastic Systems via Invariant Embedding Technique. WSEAS Transactions on Mathematics 7, 141–159 (2008) 18. Nechval, N.A., Berzins, G., Purgailis, M., Nechval, K.N., Zolova, N.: Improved Adaptive Control of Stochastic Systems. Advances in Systems Science and Applications 9, 11–20 (2009) 19. Nechval, N.A., Purgailis, M., Berzins, G., Cikste, K., Krasts, J., Nechval, K.N.: Invariant Embedding Technique and Its Applications for Improvement or Optimization of Statistical Decisions. In: Al-Begain, K., Fiems, D., Knottenbelt, W.J. (eds.) ASMTA 2010. LNCS, vol. 6148, pp. 306–320. Springer, Heidelberg (2010) 20. Nechval, N.A., Purgailis, M., Cikste, K., Berzins, G., Nechval, K.N.: Optimization of Statistical Decisions via an Invariant Embedding Technique. In: Proceedings of the World Congress on Engineering 2010. Lecture Notes in Engineering and Computer Science, London, U.K, pp. 1776–1782 (2010) 21. Nechval, N.A., Nechval, K.N., Purgailis, M., Rozevskis, U.: Improvement of Inventory Control under Parametric Uncertainty and Constraints. In: Dobnikar, A., Lotrič, U., Šter, B. (eds.) ICANNGA 2011, Part II. LNCS, vol. 6594, pp. 136–146. Springer, Heidelberg (2011)

Multi-level Fuzzy Rules-Based Analysis of Virtual Team Performance Dalia Kriksciuniene and Sandra Strigunaite Department of Informatics, Vilnius University, Muitines 8, LT-44280 Kaunas, Lithuania {dalia.kriksciuniene,sandra.strigunaite}@vukhf.lt

Abstract. The article presents model for evaluation of virtual team performance based on the intelligent methods of Multi-level fuzzy rules and Fuzzy Signature. The hierarchical system of parameters for virtual team performance evaluation is elaborated by applying expert survey. The aggregated measure of performance of virtual project team is derived from twelve parameters assigned to three categories (team, task and interaction). The experimental research is based on fuzzy analysis of interaction data of virtual teams which worked on implementation of three software solution projects. The research results provide evidence for the feasibility of using the proposed method as the tool for virtual project managers seeking to improve their leadership techniques, and to derive parameters for performance evaluation based on intelligent computing methods. Keywords: E-leadership, virtual team performance evaluation, fuzzy rules, fuzzy signature.

1

Introduction

Management activities of any project are often distinguished to project administration and leadership. This applies for project teams working in the real or virtual environment. The need for e-leadership in virtual project teams has become increasingly relevant as businesses move toward more non-traditional work [1]. The leadership of virtual work management and coordination is much more complicated not only because of the communication restrictions of the virtual working environment, but also because of inability of the manager to be aware of the moods and informal communication context within the team. Managers in virtual and other “organic contexts” are indeed less able to exercise the same influence [2]. Lee [3] indicates a gap in the body of knowledge while applying situational leadership theory to virtual project management. Konradt and Hoch [4] state that eleadership requires a different type of leadership techniques than the traditional project management. The differences of conducting real and virtual projects are especially vivid from team member collaboration perspectives. The integration and change of fast developing technologies for project / task management and collaboration processes among team members is challenging for search of effective analytical techniques which could improve insightfulness of a project leader and his ability of adequate T. Skersys et al. (Eds.): I3E 2011, IFIP AICT 353, pp. 305–318, 2011. © IFIP International Federation for Information Processing 2011

306

D. Kriksciuniene and S. Strigunaite

performance evaluation. Application of intelligent analytic techniques for processing team collaboration records can be used as the effective tool for project managers. Human interaction management theory (HIM) analyses modelling principles of the human work [5]. It emphasizes that the performance of the project work can be designed not only as the workflow of tasks and processes, but it should take into account the interaction processes among project members as well. The work models based on HIM theory should be designed from the role interaction stance by applying special notation for defining interactions related to the project tasks. By applying main concepts of this theory we research core parameters for virtual teamwork evaluation. In most cases of observing the ways the project managers evaluate teamwork situations we could state that they apply transcendental and linguistic parameter values, for example “the task is complex” or “the experience is low”. The output parameters denoting features and outcomes of the project tasks are imprecise as well. Although all project leaders apply their experience and leadership qualities for making decisions and actions in certain situations, the rules and desciptions of these situations are rather difficult to define. The efforts to define the quantitative values for the input, output parameters of the project work and their interrelationships are rather complicated, though the communication data of the project team members is extensively available from the virtual project environment. The imprecise and ambiguous characteristics of the project work imply that the methods suggested by fuzzy computational methods are appropriate for deriving necessary parameters and rules for analysis. These methods find their application in process analysis- in Fayek and Zhuo [6] the investigation of fuzzy expert system for design performance evaluation is presented. In this article we propose the method for virtual team performance evaluation by applying methods of multi-level fuzzy rules and fuzzy signature. The article is organised as follows. In section 2 the virtual team performance parameters are defined and their measurement is suggested by applying expert survey. In section 3 the application of multi-level fuzzy rule and fuzzy signature methods is substantiated for further research. In section 4 the procedures of experimental research for virtual team performance evaluation are explained, the interpretation and insights of the research results are overviewed. In the section 5 we conclude of the outcomes and present key challenges for further research.

2

Virtual Team Performance Measurement Criteria

The performance measurement criteria of virtual teamwork were researched by conducting expert survey. Possibility to extract appropriate data from team collaboration environment, definition of evaluation criteria, and their quantitative measurement problems were discussed during survey of fifteen experts. The participants of the discussion were project managers of highest experience and technical consultants of JIRA Agile and JIRA Confluent solutions. The in-depth analysis of this research is presented in [9]. By applying concepts of HIM and situational leadership theories, the survey aimed to discuss the topics: “What criteria can be used for evaluating parameters of three dimensions (task, team and interaction) characterizing project implementation in the virtual environment?" and “How these criteria can be derived from the team

Multi-level Fuzzy Rules-Based Analysis of Virtual Team Performance

307

collaboration data, registered in the virtual environment?”. According to the suggestions and insights of the experts, the hierarchical list of measurement criteria was elaborated and refined. The criteria list, their descriptions and data source types for defining their value are presented in Table 1. Team evaluation criteria are of two types (Table 1): (1) Parameters that describe team as unit, (2) Parameters for describing individual team members. In Table 1, the team evaluation criteria that characterize team as the organizational unit include its size, role variety and hierarchy level. These parameters are important for adequate evaluation, as different task/team settings require respective organisational level. The team evaluation criteria that describe individual members provide compound measures aggregated from individual evaluations of each team member. These criteria include experience and characteristics, which jointly form the team maturity criterion. Table 1. The criteria list for virtual team performance evaluation Criteria

Description

Source type

Team evaluation criteria Size

Number of performers assigned to the task during whole task implementation period

System / Value

Variety

Number of different roles assigned to task

System / Value

Experience

Level of team experience

Expert

Characteristics

Cumulative measures of personal characteristics: attitude to work/task implementation of the performer

Expert

Hierarchy

Level of team hierarchy (the rate of high, middle, junior experience within a particular role)

System / Value

Task evaluation criteria Phase

Expert judgment /manager evaluation, parameter rate from a particular interval (beginning, middle or end of a particular project phase or iteration)

Expert / System

Task intelligence level

Expert evaluation of the necessity of human driven effort, necessary to implement particular task

Expert

Result clarity

Expert / project team evaluation about quality criteria and clarity of the expected result

Expert

Interaction evaluation criteria Meeting level

Number of meetings and their type, topic, average duration

Questioning level

Number and types of questions sent to team members and requests for information to team leader and senior members

Information sharing level

Eagerness of team and individual members to share information

Punctuality level

Parameter describing team punctuality level

System/Compo und value f(t)

System / Value Yes / No set

308

D. Kriksciuniene and S. Strigunaite

Task evaluation criteria are strongly based on project specifics; therefore, they can be evaluated by applying Computations, Compromise, Experts judgement or Inspiration decision processes, suggested in [11]. As not all task implementation circumstances can be known in advance, one of the recommendations of the expert survey was to include parameters denoting task intelligence level and its difficulty. The compound characteristic of task “difficulty” is derived from the phase and result clarity criteria (Table 1). The criteria, which belong to the task category, are designed to show if implementation of particular task needs team with a high experience or high level of team interaction capabilities. The third group of criteria related to interaction evaluation describes expected interaction activeness level among team members during the specified task implementation period. The evaluation of criteria, denoting questioning and information sharing levels summarize amount of various types of messages, records in blog and wiki areas, exchanged by team members. The compound evaluation criteria, which describes the meetings held during implementation of particular task, depends on evaluation of questioning and information sharing. Punctuality shows if the interim or final results of a task are presented on time or are reported to be done within an acceptable time. All project management experts agreed that possibility to observe communication of team members in virtual surrounding is as much necessary as in real environment. The patterns of team behaviour, characteristics of communication situations can provide important insights about task implementation process. The possibilities to use virtual communication data for criteria measurement were explored by analysing the collaborative software application ComindWork [10]. In Table 1 it is shown, that the values of the parameters can be set as quantitative values (Value), computed from the appropriate data from project environment (System) or defined by project experts and leadership (Expert). In either case the values of the lowest level criteria have to be quantified. The hierarchical structure of the criteria set implies that the criteria of higher level can be computed as compound characteristics. The parameter list is designed for characterizing situation of task fulfilment by the project team. The experienced leader of the project management is expected to define the possible outcomes of task fulfilment based on evaluation of input criteria. Three output value descriptions are chosen for further investigation: (1) Well - if task implementation process is performed well; (2) Chaos (Delay) - if there is a problem of non-understanding and chaotic communication; (3) Bad - if work is delayed or stagnated. Our goal is to present method how to evaluate the hierarchical criteria, to derive rules for defining interrelationships of the input and output criteria, and assign the output values describing task fulfilment situation. The imprecise and expert-based origin of the performance evaluation criteria implies application of fuzzy methods for their further processing.

Multi-level Fuzzy Rules-Based Analysis of Virtual Team Performance

3

309

Multi-level Rule Base Method with Fuzzy Signature Approach

Multi-level fuzzy rule method and application of fuzzy signatures aggregation techniques was chosen for the design of model for evaluation virtual teamwork performance. Fuzzy signatures are used for constructing hierarchical structures of fuzzy characteristics. They can be applied for modelling the complex structure of data in a hierarchical manner (bottom up). The Fuzzy signatures can reduce complexity with slightly more complicated aggregation techniques in huge fuzzy structures. Fuzzy signature can be considered as a special kind of multi-dimensional fuzzy data, where some of the dimensions are formed as a variable sub-group and determines features of a higher level parameter or group [12]. This means that instead of assigning a single membership grade to each element X, as it is done when defining original fuzzy sets, a set of quantitative features are assigned to each element X which can have a structure of another nested vector, this way providing additional information about that specific element of the domain. This structure can be continued and finally form a signature with finite depth m [13]. The method of fuzzy signature was introduced by Koczy (1999) [13]. It is the generalization of vector valued fuzzy sets (VVFS). It means, that each element is assigned by matching A to each element x ∈X or vector component of another VVFS (branch) or atomic value (leaf) [13]. It can be specified as in formula (1): k

A : x → [ ai ]ik=1 ≡ ∏ a i ,

(1)

i =1

here

Π denotes the Cartesian product, and [ a ]ki ; if branch ai =  ij j =1 [0,1] ; if leaf

Fuzzy signature can be charted as the nested vector or hierarchical tree structure (see Fig. 1a and Fig.1b)   X 11         X 12     X 13        X 211       X 212    X =      X 213        X 21       X 31      X 32        X 33  

(a) As a vector

(b) As a tree Fig. 1. Fuzzy Signature structures

310

D. Kriksciuniene and S. Strigunaite

The whole structure of fuzzy signature looks like a tree graph. The lowest level elements of this fuzzy signature denote leaves, the middle – branches. In the Fuzzy Signature structure (Fig 1 (b)) leaves [X11 ,X12] are sub-group of the higher level compound joint X1, the leaves [X211, X212, X213] form higher level compound X21, which compounds X2 in a conjunction with X22. Therefore X2 can be expressed as [[X211, X212, X213],X22] or [X21, X22]. In the highest level of fuzzy signature it can be abstracted as X = [X1, X2, X3] or as vector showed in Fig. 1 (a). The underlying general concept of fuzzy signature is a nested vector [13]. The connections between higher and lower levels are constructed by fuzzy set aggregations. The most common aggregation operations are the maximum, minimum and arithmetic mean. The possibility to select the method of abstraction allows to modelling particular situations of performance evaluation, where some of the parameters can be assigned different weights or even lose their importance due to specific situations of task fulfilment. Fuzzy signatures are used for constructing Fuzzy Rules. In general fuzzy rule is described as the following structure: “If x is then y is ”, where is the rule antecedent and is the rule consequent, x is the observation and y is the conclusion. The rule antecedent can be a fuzzy signature set or fuzzy signature singleton. The signatures have the same arbitrary structure, and the corresponding aggregation operators are uniform for every rule. The consequent parts of the rules remain fuzzy sets [14]. The method is applied for processing the variables and criteria included to the model characterizing virtual teamwork and its performance.

4

Analysis of Multi-level Fuzzy Rules Method Application for Virtual Team Performance Situation Evaluation

4.1

Hierarchical Evaluation Criteria Structure

The results of expert survey and analysis of virtual team performance measurement criteria, as presented in section 2, were processed by applying fuzzy signature concept for designing the hierarchical virtual team performance evaluation model. The hierarchical set of criteria is presented in Table 2. The scheme refined by the participants of the survey is arranged in four levels. The lower level criteria have the lowest level of fuzziness and can be assigned values by using interaction data of the virtual space or determined by project leader. Three possible values are selected as situation output. The criteria of the higher level are computed by applying fuzzy signature method. The hierarchical virtual team performance evaluation scheme is explored by the experimental research, discussed in the following section.

Multi-level Fuzzy Rules-Based Analysis of Virtual Team Performance

311

Table 2. Hierarchical virtual team performance evaluation scheme Input criteria (leaves)

Middle criteria (branches)

(C121) Result clarity {Low; High} (C122 ) Phase {Low; High}

(C12 ) Task difficulty

Output

(C1) Task

(C11) Task intelligence level {Low; High} (C222) Characteristics {Low; High} (C221) Experience {Low; High} (C211) Size {Low; High} (C212) Role variety {Low; High} (C213) Hierarchy level {Low; High}

(C22 ) Maturity level (C2) Team

(C21) Organisational level

(C311) Meeting level {Low; High} (C312) Questioning level {Low; High}

(C31) Activity level

(C313) Information sharing level {Low; High}

(C) Situation {Bad; Chaos; Well}

(C3) Interaction level

(C32 ) Punctuality level {Low; High}

4.2

Experimental Research Setting

The goal of the experiment was to test if the suggested parameter hierarchical structures can help to identify task implementation situations. Table 3. Experimental data description Parameter

First project

Second project

Third project

Duration

2 year

6 month

1 year

Phases

9

3

6

9 – 12

2–4

12 – 20

Hierarchy

High

Low

Middle/High

Role variety

Middle / High

Low

Middle / Low

Analysed number of tasks (work packages)

144

48

480

Majority of task types

Mix

Intelligent

Difficult/Routine

Majority of task Situations identified

Mix

Well

Chaos

Working style

One big group

Individual work

Working in small / individual groups

Experience

Lack of Time

Weak Structure

Number members

of

Main reason problems

team

of

312

D. Kriksciuniene and S. Strigunaite

The data of the experimental research was derived form three IT software implementation projects, where the team members communicated in the virtual space during the most part of the projects. The logs of virtual teamwork collaboration system were used. The team leader had possibility to observe all the communication instances of the team members related to task fulfilment during the project work. All projects had teams with different experience and organizational structures, and different working styles. Brief descriptions of projects are presented in Table 3. Task was defined as a work package for certain functionality implementation. The shortest tasks lasted from two days, but majority had duration from one to two weeks. If a task was implemented on time then its outcome was marked as Well, if the task was delayed without clear outcome, then it was denoted as Chaos, and if the task planning was changed or redone, then it was marked as Bad. Limitations of the research were related to data evaluation: the virtual communication data could be used for evaluation of small part of parameters, the remaining part were expert evaluations and had high level of uncertainty. 4.3

Application of Fuzzy Signature and Fuzzy Rules

Multi-level Fuzzy rule applying Fuzzy Signatures method is suitable for this type data analysis because of the possibility to apply bottom-up fuzzy parameter evaluation and aggregation techniques. The Fuzzy Signature structure for Hierarchical virtual team performance is presented in formula (4). For parameter aggregation the arithmetic mean was used.     C   11      C   121       C122        C 211      C1    C 212    C = C 2  =   C 213      C 3    C 221      C       222      C 311        C 312      C       313      C 32  

(4)

The Fuzzy Signatures for Fuzzy Rule construction is identified by three parameters [(C1)Task, (C2)Team, (C3)Interaction]. Each parameter has several possible value intervals. These parameter value interval ranges and their clarifications are presented in Table 4.

Multi-level Fuzzy Rules-Based Analysis of Virtual Team Performance

313

Table 4. Third level parameter value intervals and their clarifications Criteria value name

Abr .

Value range

Team evaluation

Value get in these circumstances Formula :

C 2 = ((C 211 , C 212 , C 213 ), (C 221 , C 222 )) Strong team

H

[0.75; 1]

High level of Organisation (C21) and Maturity (C22) parameters.

Experienced team

MH

[0.65;0.85 ]

Lower evaluation of Organisation (C21) parameter (then structure is flat) and High evaluation of Maturity (C22).

Growing team

ML

[0.35;0.75 ]

Lower level of Maturity (C22) and High or Middle level of Organization (C21)

Weak team

L

[0; 0.45]

Task evaluation

Low level Maturity (C22) and Organization (C21). Formula:

C1 = (C11 , (C121 , C122 )) Intelligent task

H

[0.65;1]

High evaluation of Intelligence (C11) High/Middle evaluation of Difficulty (C12).

Difficult task

M

[0.4;0.7]

Evaluation of parameter Difficulty is (C12) High/Middle and Lower evaluation of Intelligence (C11).

Routine task

L

[0;0.45]

Task with evaluation of Low is defined as routine and little intelligence requiring task. This type task can be dedicated to Growing team.

Interaction evaluation

parameter

and

Formula:

C2 = ((C311 , C312 , C313 ), C32 ) Very strong activity

VH

[0.75;1]

All parameters: Meeting (C311), Questioning (C312) and Information sharing (C313) values are High.

Low sharing of information

HS L

[0.65;0.8]

Then Meeting (C311) and Questioning (C312) are High/Middle and Information sharing (C313) value is Low.

Active and proper information sharing

M

[0.45;0.7]

All parameters Meeting (C311), Questioning (C312) and Information sharing (C313) valued as High/Middle. There is balance.

Low activity but strong information sharing

LS H

[0.25;0.5]

Then Meeting (C311), Questioning (C312) parameters are evaluated as Middle/Low and Information sharing (C313) values are High.

Very low activity

VL

[0;0.3]

All parameters: Meeting (C311), Questioning (C312) and Information sharing (C313) values are Low.

Output values is parameter Situation (C) with value intervals defined as {Bad}, {Chaos} or {Well}. The ranges of each value interval is [0;0.3] - Bad, [0.3;0.7] Chaos and [0.7;1] - Well.

314

D. Kriksciuniene and S. Strigunaite

By using Team/Task/Interaction parameter interval ranges sixty different parameter combinations (fuzzy rules) are constructed and assigned to one of the possible Outcomes. Fuzzy rules are constructed by using MatLab software application. The results are plotted as the three-dimensional diagrams where the interrelationships of three categories- team, task and situation- are further interpreted. The insights are provided by analysing the fuzzy rules derived of interrelationships of input variables plotted on X and Y-axes of the diagram and output variable plotted on Z-axis. The insights on Situation outcomes by analysing Team and Task input variables are derived from the surface diagram presented in Fig.2. The patterns of various combinations are presented by using categories of fuzzy values form Table 4. Five Situation patterns are identified: • • • • •

P1-W1 Strong Team [0.8;1] reaches situation outcome which belong to the value area Well [>0.8] almost with any type of task. P2-W2 Experienced team in a range [0.7;0.8] with task type Difficult [0.4;0.6] reaches situation Well [0.8]. Noticeable that with Routine task [0;0.4] and Intelligent task [0.7;1] evaluation of situation is lower. P3-W3 Growing Team [0.4;0.5] can perform task that requires lower level of intelligence (routine type), as it was assigned Situation value [0.75] for the Task simple/routine [0;0.3]. P4-C1 Growing team [0.4;0.5] and Difficult/Intelligent [0.4;1] task lead to situation assigned to Chaos [0.4;0.5]. P5-B1 Weak Team [0;0.3] with difficult/intelligent [0.4;1] task lead to Situation value in a range [0;0.3] that means Bad. Even if low intelligence tasks can be dedicated for this team type, good situation can’t be expected.

Fig. 2. Team and Task surface diagram

Multi-level Fuzzy Rules-Based Analysis of Virtual Team Performance

315

Insights on Situation outcome by analysing Task and Interaction combination showed that mostly chaotic situations were identified. This indicates the Team parameter has the highest influence on final situation outcome. By analysing Team-Task combination results presented in the surface diagram (Fig.3) the following Situation patterns were identified: • • •

• • • • •

P6-W4 then task is Intelligent [0.8;1] and Interaction rate is High [0.9;1] Situation is evaluated as Well [0.8]. In is the best possible situation. P7-W5 if task belongs to Routine type task range [0;0.3] and Interaction rate is Low [0;0.2] the situation assigned value of Well [0.7]; P8-C2 Task is Difficult [0.4;0.8] and Interaction belongs to range [0.4;1] then Situation value is Chaos [0.5;0.6]. Analysing these type situations is possible to state that value of parameter Team plays the decisive role for defining situation outcome. P9-C3 Interaction range is in Middle [0.5] and Task is strongly intelligent [0.9;1] then situation is assigned to Chaos [0.4;0.5]. P10-C4 Task routine/Difficult [0;0.2] and Interaction denotes very low level of information sharing [0.3], in this case the sudden fall of situation assessment is noticed. P11-C5 Task Routine [0;0.3] and Interaction is High [0.9;1] then Situation is defined as strong Chaos [0.3] almost Bad; P12-B2 Task Difficult [0.3;0.6] and Interaction is Low [0.3], borderline situation assessment for Bad and Chaos ranges [0.3]. P13-B3 Interaction Low [0;0.2] and Task is Intelligent [0.9;1] then situation is defined as completely Bad [0;0.2].

Fig. 3. Task and Interaction surface diagram

Some additional insights of Situation outcomes are derived by analysing Interaction and Team combinations. Four strong patterns are identified (Fig.4 ).

316

D. Kriksciuniene and S. Strigunaite

Fig. 4. Interaction and Team surface diagram

•

• • •

P14-W6 situation within the range of [0.7;0.8] is evaluated as Well by the combination of input: rate of Interactions is middle [0.5] and Team type is Strong [0.8;1]. If Team is strong [0.95;1] and Interaction rate is High [0.95;1] situation leads to the result Well too. P15-W7 Team is experienced [0.7;0.8] then situation is identified as Well [0.7] with any value of the Interaction parameter. P16-C6 Interaction rate is Middle/High [0.4;0.9] and Team parameter belongs to range Growing [0.4], then situation is defined as Chaotic [0.4;0.6]. P17-B4 Interaction rate is Low [0;0.3] and Team is Weak [0;0.2], then situation is assigned to Bad [0;0.3].

The results presented in surface diagrams and describes as insights for project evaluation can be used for forecasting of the situation outcome and for defining strategies aimed to strengthen particular parameters in order to avoid project risk. The advantage of the presented method is its ability to derive values of the situation evaluation parameters which have the highest level of abstraction and risk from the values of lowest level parameters which can be easier evaluated by even less experienced project leaders or computed from the interaction data of virtual communication space. The methods of fuzzy signature and fuzzy rules solve the problem of the fuzziness of parameters values and their hierarchical structure. The aims of further research include refinement of criteria evaluation methods and implementation of fuzzy signature based model to the virtual teamwork environment.

5

Conclusions and Further Research

The analysis of scientific research related to project leadership techniques and their application to virtual teamwork analysis revealed lack of effective performance evaluation methods. The methods based on subjective judgement and direct observations of team members fail in the virtual environments.

Multi-level Fuzzy Rules-Based Analysis of Virtual Team Performance

317

The proposed method for evaluation of virtual team performance aimed to refine the system of parameters and apply computational method for their quantitative evaluation. The expert survey was applied for designing model consisting of twelve fuzzy parameters arranged in the hierarchical structure that characterize dimensions of Team, Task and Interaction, leading to evaluation of virtual team work performance situation. The list of parameters and can be refined by further research. The fuzziness of the input and output variables and their interrelationships led to applying the methods of fuzzy rules. The hierarchical structure of the parameter set and the “bottom-up” approach used for evaluation of higher level criteria implied using the methods and fuzzy signature. The criteria enlisted in the lower level of the suggested model can be not only evaluated by experts or project leaders, but by deriving their values from the interaction and log data stored in the virtual teamwork environments. The fuzzy rule set and their interpretation was designed by analysis of three dimensions: Task, Team and Situation. The applied principles can be refined for creation of expert database for project and task evaluation purposes with higher accuracy of recognizing problematic or chaotic virtual team work performance situations. The insights and interpretations of the research results can be useful for virtual teamwork evaluation and to reduce risk of direct evaluation of project situation by the inexperienced project leaders without taking into account the underlying lower level criteria.

References 1. Den Hartog, D.N., Keegan, A.E., Verburg, R.M.: Limits to leadership in virtual contexts. The Electronic Journal for Virtual Organizations and Networks 9 (2007); Special Issue: The Limits of Virtual Work 2. Cascio, W.F., Shurygailo, S.: E-Leadership and virtual teams. Engineering Management Review 36(1), 79–79 (2008) 3. Lee, M.R.: E-leadership for project managers: A study of situational leadership and virtual project success, Ph.D. Dissertation, Capella University, USA (2010) ISBN 9781124067209 4. Konradt, U., Hoch, J.: A work roles and leadership functions of managers in virtual teams. International Journal of E-Collaboration, 16–35 (2007) 5. Harrison-Broninski, K.: Human Interactions: The Heart and Soul of Business Process Management. Meghan-Kiffer Press (2005) ISBN 0929652444 6. Fayek, A.R., Zhuo, S.: A fuzzy expert system for design performance prediction and evaluation. Canadian Journal of Civil Engineering (2005) 7. Jarvelin, K., Wilson, T.D.: On conceptual models for information seeking and retrieval research. Information Research an International Electronic Journal 9(1) (2003) 8. Chen, S.C., Lee, A.H.I.: Performance Measurement of SMEs’ Employees by Fuzzy Multicriteria Decision Theory. In: 2010 International Symposium on Computer, Communication, Control and Automation (2010)

318

D. Kriksciuniene and S. Strigunaite

9. Kriksciuniene, D., Strigunaite, S.: Virtual Team Tasks Performance Evaluation Based on Multi-level Fuzzy Comprehensive method. In: The Third International Conference on Future Computational Technologies and Applications (in press, 2011) 10. Comind Work, Manage people online, http://www.comindwork.com 11. PMBOK. A guide to the project management body of knowledge: PMBOK guide, 3rd edn., p. 403. Four Campus Boulevard, Pennsylvania (2004) ISBN 1-930699-45-X 12. Mendis, B.S.U.: Fuzzy Signatures: Hierarchical Fuzzy Systems and Applications. Ph.D Dissertation. Australian National University, Australia (2008) 13. Koczy, L.T., Vamos, T., Biro, G.: Fuzzy Signatures. In: Proceedings of Eurofuse-SIC 1999, Hungary, pp. 210–217 (1999) 14. Tamas, K., Koczy, L.T.: Inference in Fuzzy Signature Based Model. Series Intelligentia Computatorica 1(3), 61–82 (2008)

Author Index

Arendsen, Rex

14

Botvich, Dmitri 73 Bukhsh, Faiza Allah

Mihoob, Ahmed 58 Mileviˇcien˙e, Edita 177 Molina–Jimenez, Carlos

28

Najmul Islam, A.K.M. 255, 269 Nechval, Konstantin 292 Nechval, Nicholas 292 Nemuraite, Lina 177 Nethe, Arnim 245 Novickis, Leonids 282 Ntouskas, Theodoros 44

Caetano, Artur 153, 163 Ciesielska, Malgorzata 189 Dilijonas, Darius

87

Elmisery, Ahmed M.

73

Fomin, Vladislav V.

129, 227

Haav, Hele-Mai 1 Heinonen, Olli J. 237

Pavalkis, Saulius 177 Pentafronimos, George Pereira, Carla Marques Pesonen, Tero 212 Pieterson, Willem 14 Polemi, Nineta 44 Purgailis, Maris 292

Imbrasas, Darius

Razukas, Mantas

Giannoutakis, Konstantinos N. Gudaityte, Gintare 129 Gustas, Remigijus 138 Gustiene, Prima 138

Janovs, Valdis

99 227

Kazanavicius, Egidijus 99 Koskivaara, Eija 110, 237 Kotovs, Vladimirs 282 Krebs, Irene 245 Kriksciuniene, Dalia 305 Laukkanen, Raija 237 Lesovskis, Andrejs 282 Li, Feng 189, 200 Melninkaite, Vida 129 Mian, Salman Qayyum 110

58

200

44 153

99

Sakalauskas, Virgilijus 87 Shrivastava, Santosh 58 Smolander, Kari 212 Sousa, Pedro 153 Strigunaite, Sandra 305 Teixeira, Jose Tribolet, Jos´e

110 163

van den Boer, Yvon 14 Vitkute-Adzgauskiene, Daiva Weigand, Hans

28

129