QoS Prediction of Web Services on Collaborative Filtering

With the increasing popularity of the Web services, finding desired Web services for user constitutes some of the most challenging issues of service-oriented architecture (SOA). A collaborative filtering (CF)-based recommendation system is introduced to predict the quality of service (QoS) value of Web services before selecting. However, existing approaches for QoS prediction based on CF tend to compute the weight among different CF methods depending on the experience. Our work is very different from these methods since we design a hybrid prediction method which is more appropriate for QoS prediction in weight computing. Experimental studies are conducted on a well known real-world Web service performance dataset consisting of 1.5 millions invocation results to demonstrate that our approach provides better prediction results than other approaches.