A reputation model based on hierarchical bayesian estimation for Web services

The motivation of Web service comes from its interoperational ability so a large number of Web services can interact with others and constitute an open network, Web service network. The success of Web services selection rely on, not only its Qos capability advertised, but the trustworthy of QoS to large degree. How to evaluate the trustworthy of services QoS information, however, is a challenge in Web service network. Reputation system, a mechanism which assesses the future QoS performance by the past behavior of service, is one of promising approaches to facilitate users make optimal decision. In this paper, we present a hybrid framework of reputation model for Web service. Based on this hybrid architecture, clients build their specific social communities, by which they obtain service´s prior reputation. At the same time, the central reputation system fuses the rating data from clients by bayesian estimation. The result of experiments illustrated our approach is more efficiency and accuracy in several aspects, especially when dealing with strategics services.