Web Service Composition Optimization By Integrating QoS And Credibility Computing

Web service is a typical technology of Service Oriented Architecture, a single Web service has already cannot satisfy the user's complicated request because of the limited functionality. Therefore, composing existing simple Web services to build complex value-added services gets wide attention and a large number of technologies about Web service composition have emerged. Among them, QoS-aware Web service composition is becoming the focus of current research. That is, on the premise of system's functional requirements have been met, according to the non-functional attributes (Quality of Service, QoS) to find the optimal combination scheme.However, in the open Internet environment, a variety of offered services are all in a dynamic process, there will be fraud more or less due to the lack of regulation, high QoS derived from traditional service selection approach cannot guarantee the credibility of service, result in the final composite service cannot guarantee the satisfaction of user. In order to solve above problems, this thesis proposes a QoS-aware Web service combination optimization method combined with credibility computing.Specifically, this thesis gives credibility evaluation and weight calculation for Web service's QoS attributes from objective aspect and subjective aspect. Bayesian learning theory and users' evaluation are used to evaluate reputation and trust respectively, and a weight calculation method is given based on rough set theory and fuzzy analytic hierarchy process. So that a QoS measurement model integrated of credibility is put forward. In local optimization phase, filtrate candidate services according to the results of credible QoS measurement, and then the selected credible services are applied in subsequent global combinatorial optimization. In the stage of global optimization, the service composition problem is established by multi-objective optimization model, and is solved by an improved cuckoo multi-objective search algorithm, aiming to get the global optimal composite services. Finally, a series of experiments are conducted to assess and verify these methods. Experimental results show that the proposed service selection method can not only satisfy the requirement of QoS but also ensure the credibility, and the performance of our proposed improved algorithm is significantly better than other several classical algorithms.