Research And Implementation Of Web Service Recommendation Based On QoS Prediction Algorithm

With the wide application of SOA, the number of Web services on the Internet is increasing rapidly, which makes the search, select and invoke Web services no longer meet the needs of the users of Web services. Thus, the efficient discovery, recommendation, selection and combination of services algorithms become an urgent research needs. How to find the service that meets the needs of the user’s functional requirements, is the research content of the service discovery. However, in recent years, there are many function-equivalent services with different QoS (quality of service) on the Internet and the cloud platforms. It’s not enough to find the service which meets the needs of the service How to select some of the more appropriate services according to their non functional properties from a large number of functionally equivalent services becomes a hot and difficult research topic.However, the QoS of a service has shown a strong uncertainty to different users who in different network environments. Therefore, according to the static QoS provided by service provider or average of historical QoS to select service, the users may not be satisfied with the results. In addition, because of the continuous surge in the number of Web services, and even show explosive growth trend, the QoS of the vast majority of Web services is unknown to the users, that is, the incompleteness of Web services QoS. The uncertainty and the incompleteness of the QoS, makes the service selection based on QoS has great difficulty. Therefore, it is very important to predict the QoS of the service that the user has not used based on the existing QoS information.Based on this, this paper proposes a manufacturing neighbor collaborative filtering prediction algorithm for Web service QoS. Firstly, the algorithm is based on the context information filtering to predict the QoS of the stable users and the stable of Web services. Then, the iterative thinking is introduced, and the collaborative filtering prediction algorithm is used to predict the QoS of the ordinary users and ordinary services. Finally, predicts the QoS which has no historical QoS and is not predicted. The experimental results based on a real public dataset show that the algorithm has better prediction accuracy.