Research On Web Service Discovery Recommendation Algorithm Based On Service Mining

Now, with the rapidly development of network technology, especially with the increasing of Web services, how to quickly, accurately and efficiently find services which satisfy the user requirements from a large mount of Web services has become a serious problem. In recent years, automatic discovery Web services is become the research hot topic, which attracts so many researcher's attention.Traditional service discovery methods are mainly by matching keywords at the syntactic level of services, these methods can not identify the semantic information well, resulting in its precision and recall are relatively low. The existing semantic-based service discovery methods denote their service functions by input parameters and output parameters, however, the precision results depend on purely functional matching are not high.With all these problems in mind, we propose a service discovery method based on clustering service groups, meanwhile, a service recommendation method based on QoS constraint is put forward. The group-based Web service discovery mining method is based on the basic idea of the collection of Web service logs, constructs a cross-operating graph through interaction frequency and behavior similarity between operations, obtains the similar matrix of service operation point by computation. Further, the proposed method clusters the interaction graph into several different internal frequent interactive sub-graphs by using spectral clustering methods and K-means method. Each sub-graph present a cluster which work together to complete the same business goals From a business point of view, thereby reducing the service search space to a large extent and find the services to meet user needs quickly and effectively. On the other hand, because of a large amount of the same or similar functions become available by being published and registered as Web services, quality of Service QoS is a major challenge in terms of discovering the correct services. Further more, we propose a service recommended model based on QoS constraint by analyzing and evaluating several factors which impact QoS quality. All in all, the recommendation model not only quickly finds Web services to meet the needs of users but also ensures the recommendation accuracy of service users. The thesis designs simulation execution environment, realizes the proposed algorithms and implements tests on the simulate service sets.