Credibility Assessment Web Service Composition

With the rapid development of the Internet, web services as a form of software, its applications become more widely.And a single web service can not meet the increasingly complex needs of users. The web service composition can solve this problem effectively.With the emergence of similar functional web service, how to choose a trusted web service composition has become a hot issue. The choice of services depends on the evaluation of the web service composition.Therefore, this paper mian researchs the trustworthy of web service,and aims to provide trusted web service composition evaluation methods.This paper main includes the following three research content.1) In order to resolve the domain dependence issues, it proposes a trusted measurement model of web service based on domain ontology. First, it builds the basic and domain trustworthy attributes of web service based on basic and domain features. And it uses the domain ontology to semi-automatic build the domain trustworthy attributes. Then it formally describes the trustworthy attributes model by the ontology semantics.Finally, it proposes classify method, weights calculate method and metric method.2) To resolve the user preferences issues, it proposes a trusted evaluation method for atom-web-service based on preferences recommended. By comparing user preference similarity, it finds the evaluations of the subjective trustworthy attributes, which meet the needs of user preferences.Then, combining evaluation information of objective trustworthy attributes, trusted evaluation value of atom web services calculate can be calculated.3) It proposes a global-based personalized evaluation method for web service composition to resolve the global and personalized issues. It dynamically adjusts the evaluation model with a global point of view.Then using user attention to express the atom-web-service’s influence on trustworthy of service composition. With the earlier studies, a global-based personalized trusted evaluation model is built. It also analyses the application of the model. Finally, using simulation experiments to verify the proposed methods and the results show that the methods are trusted.