Research On Service Evaluation Method Based On Collaborative Filtering

With the development of services, more and more services are published on the network by different developers. More and more services' function are similar. In order to satisfy users' need, as the foundation of service composition and service discovery, service evaluation becomes a hot research.In order to feedback more satisfactory services to the user. Service evaluation calculates the quality of service through a set of evaluation criteria. Traditional service discovery methods do not provide the corresponding service evaluation methods. Users select their demanding services according to their own experiences. As the increasing of services, service evaluation method is demanded more and more by users. So a lot of service evaluation models are emerged to evaluate services. But for the different users and different services, the existing service evaluation methods use unified model to evaluate service, whereas, different users use the demanding services in different areas. They may have their own preferences which are not reflected in traditional service evaluation methods. In order to solve above problem, this thesis proposes a CFSEM service evaluation method to evaluate the services. Through analyzing the traditional collaborative filtering methods, this thesis introduces the user-item based collaborative filtering algorithm , achieves evaluation forecast of the service evaluation factor firstly and uses the entropy method to calculate the weight of the service evaluation factors in the services comprehensive evaluation . The services is ranked and recommended to the users according to the service evaluation value which is the weight sum of the service evaluation factors.Finally, this thesis designs a service recommendation system. This system can mine the services that aren't evaluated by users, namely, forecast the services evaluation value which aren't given by users. The about the service evaluation method is done. This experimentation proves that the method is feasible.