QoS Prediction Of Web Services Based On Collaborative Filtering

With the rapid development of Internet technology,the number of similarly-featured Web services has proliferated.How to choose the right service for users from many similar Web services is one of the hot research fields.QoS(Quality of Service)refers to the non-functional attributes of Web services.QoS prediction plays a crucial role in Web service selection and recommendation.Collaborative filtering algorithm is one of the commonly used algorithms for Web service quality prediction.However,the traditional collaborative filtering method uses single attribute data,only considering the quality of service,and does not take into account the additional attributes of users and services,These features have great value and significance for improving the accuracy of Web service QoS prediction.In real life,the number of Web services is increasing,and users cannot access all Web services to obtain their service quality.Therefore,the user-service matrix is sparse,which seriously affects the prediction accuracy of service quality.In order to solve these two problems,two algorithms are proposed in this paper:In order to make full use of the characteristic attributes of users and services and improve the prediction accuracy of service quality,a collaborative filtering prediction method RBCF based on RBF neural network is proposed.This method takes into account the Feature attributes of users and services,such as the longitude and latitude and the country where users and services are located,based on feature attributes,RBF neural network prediction model is used to predict all missing QoS values of Web services,which filling the original sparse user-service matrix data,and combines the collaborative filtering algorithm to obtain the final Qos prediction value of Web services,improving the accuracy of Web service quality prediction.Aiming at the sparse problem of user-service matrix caused by the lack of historical QoS data,only making full use of the data in limited data sets can improve the accuracy of QoS prediction.In this paper,an improved hybrid collaborative filtering based method for Web service QoS prediction is proposed,In this method,service forward prediction,service reverse prediction and similar neighbor cross prediction are introduced.User-based collaborative filtering,service-based forward and reverse neighbor collaborative filtering prediction,and cross prediction algorithm based on similar users accessing to similar services are fused by balancing factors,Using their respective confidence levels as fusion weights,The confidence can be used to measurethe credibility of the results of different prediction algorithms.Through the experimental observation and analysis,it can be seen that the improved algorithm improves the efficiency of data use,and improves the accuracy of Web service quality prediction.