Research on the Clustering and Composition of P2P-Based Web Services

For the problem that the centralized mechanism for semantic service discovery is still difficult for realization in the server center during service discovery and matching, a novel model of web services is proposed which makes use of some ideas from the peer-to-peer structure and the clustering methods. In this model, a service is firstly matched within its located community in accordance with the semantic information for its function, and then a rather accurate matching is made according to the semantic information from input parameters. Both theoretical analysis and simulation results show that this model is able to raise the recall ratio as well as to improve the matching speed, and moreover, it can also provide users with value-added response from web services composition which may meet user´s further requirements. These features not only satisfy some user´s potential needs, but also deal with the problem of the centralized mechanism for semantic service discovery during the serviceoriented computation.