Research On Web Service Recommendation Based On QoS And Collaborative Filtering

With the rapid development of Internet technology and the economy,more and more Web services are deployed on the Internet,and many Web service providers provide a large number of similar functions of Web services.It is difficult for users to choose a Web service which meet the needs of users of Web services through only functional description.The recommendation system is an information filtering system which can effectively solve the problem of information overload.Its application in Web service recommendation field can help users choose QoS to meet the needs of Web services.Its main process is: by mining the user QoS records,to predict the active user missing QoS value,and according to the predicted QoS value to select the better quality of service Web services recommended to the active user.However,in large-scale sparse data,QoS value prediction accuracy is relatively poor.Aiming at this problem,this paper proposes a context-aware Web service recommendation model based on factorization machine and a Web service recommendation algorithm based on user geographic location and user similarity.(1)The context-aware Web service recommendation model based on factorization machine analyzes the influence of context information on Qo S value,extracts the relevant context information as the characteristic of QoS evaluation,and uses the factorization machine model to predict the user's missing QoS value.By comparing a series of experiments on the real data set,it is proved that the Web service recommendation model has high accuracy on large-scale data set.(2)The Web service recommendation algorithm based on user geographic location and user similarity analysis of the user's geographical location and similar neighbor users on the quality of service impact,according to the real distance between users and the similarity between users to build two users recently Neighbors,and add them to the matrix decomposition model to predict missing QoS values.Through a series of comparative tests on the real data set,it is proved that the algorithm has high accuracy.