Research On Web Service Composition Method Based On Real-time QoS Prediction

With the rapid rise of cloud application platforms,the number of Web services is exploding.Web services deployed in the cloud environment provide a variety of services for users,and a large number of candidate services appear.The value of QoS that important index to measure the performance of Web services can only be collected when its web service was invoked.Therefore,cloud platform should get the value of QoS in advance to provide suitable Web services for users.Because of the rise of online applications(e.g.e-commerce,travel APP),cloud application platforms provide users with composite services and meet users' preferences and quality of Web service.Therefore,the construction of Web service composition has become research content.This paper has done the following research work:1.The user and service geographic information is used to cluster users and servcies,and the category information of users and services is intergrated into the matrix factorization algorithm.The addition of user categories and service categories can compensate for the problem that the matrix factorization algorithm ignores the similarity between users and services.In this paper,K-means clustering algorithm is used to cluster users and Web services,obtaining the user categories and service categories.According to the similarity between group of categories construct Regularized items that integrate real-time matrix factorization model.It can improve predicting accuracy and solve the cold-start problem to some extent.This algorithm is tested on WS-Dream dataset,and the results show that this algorithm can improve the predicting accuracy and solve the cold-start problem to some extent.2.Modeling Web service compositions with multiple objectives.Firstly,the preferences of user are categorized into qualitative preference and quantitative preference.Then,the first sub-goal is calculated under qualitative preference,the second sub-goal is calculated under quantitative preference.And,the reliability index of single Web services is extracted,the global reliability of composite services is calculated,and the third sub-goal is calculated under global reliability.Finally,the multi-objective optimization algorithm is used to calculate the final optimal composite service.The method is tested on QWS dataset and the results showed that the method improved the efficiency of the service composition.