Towards Web Services Interaction Mining Architecture for e-commerce applications analysis ROBERT GOMBOTZ 1 , KARIM BAÏNA 2 , and SCHAHRAM DUSTDAR 1 1 Distributed Systems Group, Vienna University of Technology, AUSTRIA 2 ENSIAS, Université Mohammed V - Souissi, B.P. 713 Agdal Rabat, MOROCCO r.gombotz@infosys.tuwien.ac.at, baina@ensias.ma, dustdar@infosys.tuwien.ac.at Abstract: - Nowadays, Web service technology plays an increasing role in internet applications, in general, and e-commerce applications, in particular. In fact, service-oriented systems can be expected to grow larger in complexity. Such large systems demand for tools that allow for analyzing and monitoring of service-oriented systems in use. Our work attempts to draw the necessary architecture in order to analyze interactions between Web service consumer and provider. WSIM modelling architecture is built over three layers : Web service operations, interactions and workflows. The paper aim is to present WSIM modelling architecture and how it could be implemented to support existing Web service applications (e.g. e-commerce applications). Key-words: e-Commerce Services, Web service interactions, Web service logging, Web service mining 1 Introduction It becomes obvious that Web service (a.k.a. WS) technology will be indispensable in building and integrating internet applications (e.g. CRM, SCM, e-commerce applications) [1]. Those service-based applications are expected to grow larger in complexity. In order to make WS applications easier to use and maintain for providers and customers, interesting challenges are highlighted: o Discovering complex patterns within Web Services applications (e.g. identifying reconfigurable web service architectural patterns, runtime web service behavioural patterns) o Supervising and monitoring of Web Services applications (e.g. by building analysis and administration scoreboards for web-service based applications) Within this problematic, our work propose modelling and implementation architectures for Web Service interactions analysis. Mining describes the process of discovering knowledge in large amounts of data (in our case Web Service applications data). Our contribution is not specific to a mining technique; it proposes generic architectural bricks so using of specific mining technique could be always possible. The remainder of this paper is structured as follows. Firstly, we present our WSIM modelling and implementation architecture, and finally we discuss related works before concluding. 2 Web Services Interaction Mining – Modelling Architecture We develop our Web Services Interaction Mining (a.k.a. WSIM) approach with regards to three levels of abstraction that represent three complementary Web services ”views”. Figure 2 depicts a stack of views on Web services. As one goes from the top to the bottom, the level of abstraction falls and we are looking at things in higher detail. Fig. 2. Web Service Mining Levels In the following subsections we will detail each of the three WSIM architecture levels. Knowledge Discovery Web services Workflows Web Services Interactions Web Services Operations