Research Of Web Service Recommendation System Based On Heterogeneous Network Clustering

With the rapid development of web technology, the Web Service was proposed and widely applied as a new reusable software function. Since the Web Service is flourish and personalized demand of service requester is increasingly protruding, traditional service discovery and selection mode is difficult to satisfy the user’s request. Service recommendation dedicates to provide web services which meets users’ personalized needs for the users, and becomes a new research hotspot. Among them, the Web Service Clustering clusters web services with the same or similar features, and reach the aim of improving the service recommendation effects.Existing Web Service Clustering methods are mainly based on Web Service description information, and ignore the relationship between each participants of Web Service. Aiming at this problem, this paper studies Web Service Clustering method based on heterogeneous service network, makes improvement to the classic Net Clus algorithm, improves Service Clustering effect by using the relationship between Web Service and each participants. According to Web Service Clustering information, Web Service recommendation system is analyzed and designed, by implementing the prototype of system, the algorithm is applied to the Web Service recommendation system.The main work of this paper is following:① Through looking at the domestic and foreign journals and reading relevant materials, the author analyses the main technology and standards of the Web Service, and characteristics of the existing various clustering methods; searches the development process of the information network analysis method and the present situation of the collaborative filtering recommendation algorithm.② A heterogeneous service network is constructed based on the relationship between Web Service and various participants by the way of heterogeneous information network analysis method. The relationship between Web Service and Web Service attribute, Web Service and provider, Web Service and user are considered in this heterogeneous service network model, and introduce the response time into the model which provides the model basis of web services clustering method based on the heterogeneous network.③ This paper proposes MAO-Netclus algorithm. It solves the problem that a single target node cannot connect with multiple nodes in one attribute node in the original Netclus algorithm. The simulation experiments verify the improvement performance of the clustering effect of the improved algorithm to the original algorithm, insure the improved algorithm stability in different sparseness, cluster numbers, and iteration times.④ The paper analyses the shortages of the existing Web Service recommendation systems, and makes a detail description of the Web Service recommendation system requirements based on these analysis. The author designs system architecture, function structure, and database by requirement analysis, and designs the specific process of each function module in details.⑤ The author implements the Web Service Recommendation system prototype using Java and My SQL according to the designed model of the Web Service Recommendation system. The pages of various functions in implemented system are presented, and function tests are made to this recommendation system at last.