Web Services Reputation Evaluation Model Based On Qos And User Recommendation

In 1999, Microsoft put forward "Web Services" in a press conference first time. Until recent years, Web Services are general concerned by people with its interoperability and versatility. But two Web Services companies, Microsoft and IBM, only provide development platform and communication protocol in the technical aspect. They don't give the appropriate solution on Web Services recommendation and Web Services selection. Therefore, a Web services reputation evaluation model based on QoS and user recommendation is presented.First of all, research results of Web Services reputation evaluation are summarized, ideas and features of existing model are analyzed, and their advantages and shortcomings are concluded.Secondly, to solve the problem that QoS lacks of credibility, the Web Services Concept Model is extended, and a service monitor center is designed. A QoS-updating mechanism is proposed, and QoS is updated by weighed summing provided-QoS and collected-QoS. The QoS credibility can be improved by this QoS-updating mechanism.Thirdly, the concepts of preference-similar user, recommendation user, rating reliability etc. are given. A service reputation evaluation model based on QoS and user recommendation is presented. Firstly, the QoS comprehensive index is calculated according to the QoS calculating model. Then, a group of users who have similar rating are selected by Pearson correlation coefficient, and the recommendation of the service is calculated by the index of rating similarity and the user rating. By combining the two values, the reputation of service is calculated for selecting the best service. Moreover, for the problems of malicious recommendation, a method is given to calculate the user's rating reliability.Finally, based on QoS-update mechanism, service reputation evaluation model and collaborative filtering technology, a service recommendation platform is developed for service recommendation, and a testing tool is provided for users. This platform shows the feasibility and availability of this model.