Research on web services composition

: selection, verification, and perception

Abstract

Web services are an emerging and promising technology in distributed applications. Due to rapidly increasing number of available web services over the Internet, it is necessary to aggregate several existing services to build new value-added services to fulfill complex requirements in an efficient and cost-effective way. Meanwhile, services are often long running, loosely coupled, and cross administrative boundaries, so it is also necessary to adapt transaction technique in services composition. Moreover, in order to make service selection reasonable in an open environment, reputation has been regarded as one of the most important mechanism to identify good services from bad ones. However, a composite service only obtains one perception from the user after every invocation. In order to compute the reputation of each component service, it is necessary for the composite service to distribute this perception to its component services recursively. Therefore, how to select a batch of component services to construct a composite service to fulfill complex requirements, to verify transactional properties to ensure a composite service can always terminate correctly, and to propagate the perception from the composite service to its component services fairly, becomes challenging and significant problems in web services composition. In this dissertation, we present a web services composition framework to address the problems of selection, verification, and perception by incorporating various techniques, including constructing skyline composite services, validating transactional properties of BPEL-composition, and propagating reputation from composite service to its component services. The main research works and contributions of this dissertation include the following: • We propose a composition approach to construct all skyline candidate composite services for each individual request. We also use a cluster-based grouping method to improve efficiency of constructing a set of skyline candidate solutions. We evaluate the efficiency and effectiveness of our methods via synthetically generated datasets. • We present a technique to analyze BPEL process and other related files to validate whether this process complies with the expected transactional properties. We identify key points and risk points to help validate whether a process is wellformed. The validation technique can assess the correctness of BPEL process specifications and therefore can help process designers to identify ill-formed processes. • We propose a fair perception score distribution framework which combines the context of component services and their runtime performance. We distinguish two aspects of contexts of a component service, i.e., structure-related importance and community-related replaceability, and compute them by graph theory and dominating relationship technique, respectively. Experimental results show that our approach can achieve a more reasonable and fair perception score distribution than other existing methods.

Date of Award	3 Oct 2012
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Qing LI (Supervisor)