Research On Web Service Recommendation Algorithms Based On Collaborative Filtering And QoS Prediction

Web Service, as a widely used distributed computing model, has gotten focus attention from academia and industry in recent years. However, as the expansion of Web Service scale, a major challenge has occurred in service computing field is that how to choose the optimal Web Service for the user. A good Web Service recommendation system, on one hand can promote users’ retrieval efficiency, on other hand can also help service providers to prevent losing users due to information overload.In order to help users choose the most suitable Web Service, in a group of Web Services which is functionally similar, Quality of Service(Qo S) attribute of Web Service plays the key role. Currently, the research on the Qo S-based Web Service recommendation system is limited, facing the issues of less personalized level and lower recommendation efficiency at this stage. In this paper, aiming at solving the main problem Web Service recommendation system faced, the personalized collaborative filtering recommendation algorithm within it has been explored and researched.The paper’s main work is as follows:① Analyzes the appearance background and its current situation of Web Service recommendation and recommendation system in recent years; Describes relevant technology such as collaborative filtering, clustering; emphatically introduces collaborative filtering-based Qo S prediction algorithm and user similarity calculation method.② Based on the research of user similarity in UBCF algorithm, brings in the service similarity impact factor to improve calculation method on traditional user similarity, which method combines with the user-based collaborative filtering algorithm; Proposes an adaptive service collaborative filtering Qo S prediction algorithm(SA-UBCF).③ Through the research on cluster-based collaborative filtering algorithm, brings in the concept of user confidence to solve the grey sheep problem encountered during clustering process; Proposes a user confidence-based quadratic clustering algorithm to classify the user; then applies it to recommendation algorithm, proposes a quadratic clustering-based collaborative filtering Qo S prediction algorithm(DCCF).④ Adopted the Qo SDataset2 version of WSRec dataset, experimentally evaluates the proposed SA-UBCF algorithm and DCCF algorithm, to verify its reasonableness and effectiveness.Experimental results show that comparing with the traditional UBCF algorithm and other improved algorithms which were presented by other papers, SA-UBCF and DCCF algorithms have a certain degree of improvement in the accuracy of Qo S prediction, and DCCF algorithm can also improve system efficiency recommendation effectively.