Reasearch On Trust-Based Web Service Recommendation And Discovery

Service-Oriented Computing (SOC), as a new Computing mode, takes the service as basic components to support fast and low-cost application composition of distributed and heterogeneous environment. With the continuous development of SOC, there is a remarkable rise in the types and quantity of service. Web services support for collaborative operation between different software, and also an new online evolution form of distributed software system. Nowadays, it has experienced a rapid growth on the internet. Therefore, facing the vast amounts of Web services, the problem of how to choose reliable and available ones that meet users’requirements arises.As a kind of important support, the technology of Web service recommendation and discovery solved this difficult problem. Most of the existing service recommendation technology is based on the similarity between users and the Web service itself to recommend, ignoring the trust characteristics between users, and facing the challenges of low-accuracy and small-coverage. In addition, most of the existing service discovery technology is from the perspective of user needs and Web service keyword matching and lacks consideration of dynamic change in network environment, which may cause services that user selected can’t work normally, just a problem of service failure. It affects the credibility of Web service to a certain extent. To address the problem, this paper tries to improve Web service credibility with the trust theory. The main content are as following:Firstly, this paper proposes a method of trust-based Web service recommendation (TBWSR). Combining traditional collaborative filtering algorithms and knowledge of trust network, this method recommends services to users, which it based on the evaluation information between user and Web service and trust information between users. This method takes the similarity and trust of user into consideration, thus effectively easing the negative effect brought by low-accuracy and small-coverage. In addition, the method ensures the credibility of Web services that recommended, and improves the performance of recommendation.Secondly, aiming at the problems that the Web service recommended by TBWSR method could not be invoked and lead to the decrease of its credibility, the paper puts forward discovery method of Web service in the service failure context (DWSFC). It has found the quality of Web service description file is vital for its availability. However, the exiting web service discovery technology didn’t take it into account. Therefore, this paper adopts the method based on the similarity of service function and quality evaluation of service description document, then chooses an available web service of similar function and high quality to replace the failure one. The method is presented based on particular context of the recommended results above, and it guarantees the availability of Web service and effectively optimizes the recommended results.Finally, in order to verify the feasibility and effectiveness of TBWSR and DWSFC method, some experiments are conducted and a prototype platform is designed and developed.