Research On Web Service Community Construction Method

In order to manage Web service resources effectively, Web service community comes up. Web service community is defined as the collection of Web services with the same functional property and different non-functional ones. With the increasing complex business processes and reused components, Web service community is applied in the range of both atomic and large-granularity Web services. Service discovery and replacement based on service community can improve the efficiency of service composition. The traditional Web service community construction method is implemented through users’manual registration, which has a low efficiency and is difficult to organize and manage service resources effectively. Therefore, how to construct Web service community automatically has become an important aspect in the research of service discovery.For the above issues, this thesis proposes a Web service community multi-layer model. The model contains two layers of the atomic Web service community and large-granularity one. During the process of Web service community construction, based on some methods related to complex network, this thesis proposes an atomic Web service community construction method based on weighted GN algorithm and a large-granularity Web service community construction method based on DW-Newman algorithm. The weighted GN algorithm makes similar atomic Web services in the same community by mining the community in similar relation-based Web service complex network. And the DW-Newman algorithm mines the community in invokable relation-based Web service complex network and extracts the paths between certain two nodes to discover large-granularity Web services. The method makes large-granularity Web services which have similar functions or belong to the same business area in the same community. For the proposed methods, this thesis carries out the numerous correlation experiments based on the dataset of China Web Service Cup. Besides constrccing Web service communities, the community atomic service similarity model and the community large-granularity one are proposed to analyze the rationality of construction results.The experimental results show that compared with the traditional community detection algorithms of complex network, the proposed methods in this thesis are more correct and effective. Atomic services are divided as functions with the atomic Web service community. When an atomic Web service contained in large-granularity one fails, it can be replaced by another similar service, which can improve the stability of the large-granularity service. In addition, large-granularity services are divided as business areas with the large-granularity Web service community to improve the efficiency of service discovery. Thus, the Web service community construction methods proposed is meaningful to Web service discovery and selection.