Research On Prediction Method Of Web Services Quality Based On QoS

In recent years,the rapid development of service-oriented architecture(SOA)has led to the explosive growth of the number of Web services,resulting in the emergence of a large number of Web services with the same or similar functional attributes,so users began to focus on the non functional attributes of services.In general,quality of service(QoS)can be used to represent the non functional properties of services.In real life,the QoS attribute matrix of user service relationship is often very sparse,so the accurate prediction of the QoS attribute value of web service becomes an important research direction in the field of service computing.Collaborative filtering technology has become an important means in the field of service quality prediction,including memory based collaborative filtering technology and model-based collaborative filtering technology.The collaborative filtering technology based on memory mainly predicts QoS by finding similar users of target users or similar services of target services;the collaborative filtering technology based on model mainly uses machine learning and other technologies to build models and use models to predict.In view of the above two classifications,the paper proposes the service quality prediction method based on improved collaborative filtering and the service quality prediction method based on two-stage clustering of SOM and k-means.In view of the traditional Pearson correlation coefficient is greatly affected by the sparse matrix,this paper improves the similarity calculation method,introduces the similarity weight coefficient,and combines the jacquard similarity and the weighted Pearson correlation coefficient to effectively improve the performance of the traditional Pearson correlation coefficient when the matrix is very sparse.At the same time,considering that there may be range abnormal data and malicious evaluation data in user feedback,we filter out untrusted users by a user trust calculation method.Compared with the traditional prediction method based on Pearson similarity,it improves the accuracy of QoS prediction effectively.For the traditional K-means algorithm is greatly influenced by the initial center point and the number of clusters,a two-stage clustering method is proposed in this paper.In the first stage,SOM neural network is used to train the number of clusters and the center of clusters,which is regarded as the clustering center of K-means algorithm.In the second stage,K-means algorithm is used to further cluster,which can improve the clustering effect and get accurate clustering information.In this paper,the two-stage clustering algorithm is used to cluster users and services respectively,and the Top-k selection mechanism is used to get similar user set and similar service set.The prediction value based on similar users and similar service is combined to make mixed prediction.Compared with the K-means clustering method and the classical singular value decomposition method,experiments on real datasets can effectively improve the accuracy of service quality prediction.