Reusing process fragments for fast service composition: a clustering-based approach. Issue 1 (2nd January 2019)

• Record Type: Journal Article
• Authors: Mezni, Haithem; Kbekbi, Mariem
• Is Part Of: Enterprise information systems. Volume 13:Issue 1(2019)
• Page Start: 34
• Page End: 62
• Publication Date: 2019-01-02
• Subjects: Web service composition -- service process fragment -- fragment reuse -- quality of fragment -- formal concept analysis; Information storage and retrieval systems -- Periodicals; Management information systems -- Periodicals; Electronic commerce -- Periodicals; 658.4038011
• DOI: 10.1080/17517575.2018.1460866
• Languages: English
• ISSNs: 1751-7575

Abstract

ABSTRACT: With the proliferation of Web services over the Internet and due to the increasing complexity of users' needs, Web service composition has emerged as a powerful method of software reuse, allowing to deliver complete business processes as a set of interacting services. To guarantee a rapid and secure service composition, fragments of available business processes at different granularities may be considered as a composition unit and recombined to deliver effective compositions. Despite the benefits of this method, most of the existing works do not take into consideration the reuse of service process fragments (SPF). Reusing SPFs allows, not only to minimize the composition time, but also to improve the reliability of the composition process. In this paper, we propose a Web service composition approach that aims to combine service process fragments rather that atomic services. We adopt a powerful mathematical model called Formal Concept Analysis (FCA) to define the relationships between services and fragments. Moreover, we exploit the regrouping capabilities of FCA by proposing algorithms for the extraction of candidate fragments' combinations. A scoring function is also defined to determine the quality level of each SPF and its ability to participate in a composition. The experimental studies proved the effectiveness of our FCA-based approach compared to existing state-of-the-art solutions.