Research On Technologies Of Web Services Discovery Based On Semantics

With the rapid development of Internet, there are more and more Web services available on the Web. In order to make use of these services efficiently, the Web services must be supported by appropriate discovery mechanism. In general, Web services discovery technology utilizes WSDL and UDDI, which based on keywords and classification to discovery services. However, this method can't help the users to discovery expected services and has low precision and recall.In this paper, the author proposes a Web services discovery mechanism based on services filtrating and semantics resulted from the existing services discovery mechanism. The services filtrating mechanism in this article describes Web services by tree structure, constructs latent semantic space based on advertising services from UDDI center by latent semantic analysis mechanism, makes advertising services map into advertising index vectors and constructs a database about these vectors, makes the request services map into request index vectors and calculates similarity of request and advertising index vectors then filtrates the fit services according to similarity. After services filtrating, this author utilizes a semantics-based Web services matching mechanism to match the filtrated Web services deeply, this mechanism describes the Web services as services descriptions trees with OWL-S that added the quality of services in, gives different weights to services attributes nodes according to different layers and stores them by classification, then matches the Web services by using an assistant technology and a semantic similarity calculation method based on WordNet and HowNet.In this paper, the author builds a prototype based on the Web services discovery mechanism mentioned above to prove the validity of this mechanism. The experiments prove that this Web services filtrating mechanism based on latent semantic analysis has higher filtrating precision and recall, can filtrate the majority irrelevant Web services for the further Web services matching, reduces the services matching range, saves the services matching time and improves the services matching efficiency. The Web services matching mechanism proposed in this paper is fit for matching larger range and more classes Web services than the classical services matching mechanism augment UDDI Registry system, at the same time the mechanism has higher precision, recall and stability, improves the services discovery efficiency and performance.