Research On Key Technology Of Semantic Web Service

With the development of Web Service and E-Commerce, there is an in-deep study about the high dynamic, robustness, high performance, security of Web Service, the semantic Web Service and automatic composition of Web Service, and so an ,both at home and abroad. Among them, the research on publication and discovery mechanism of Web Service is one of the focuses of concern. As one of the key of the Web Services, the discovery mechanism of Web services is very important to fulfill the challenges in the Web Services, and UDDI is one of the solutions in the present service discovery. There exist two main shortcomings in the traditional technology of the Web Service discovery: firstly, the keyword searching based on business name, service name and tModel is just supported by UDDI; semantic searching is not supported by the registry; and the service matching in the current system is just syntax based(the string based). Secondly, UDDI is a centralized structure. The current UDDI registry adopted a single registry center framework like telephone directory. However, it will bring about network bottlenecks and single point mistakes at the information explosion time. It can not resolve the search to private registry and the isolated island problem with a relatively low expansion. What is more, it is easy to be attacked with static IP and it is also difficult to make service composition and interaction automatically. Based on the shortcomings of UDDI, the structure of UDDI, semantic description of Web Service and the relative P2P technology and the semantically increased UDDI are put forward based on the previous investigations and our innovation in this thesis. The detailed studies are as followings.Firstly, we join the UDDI peers which are from different area and can form a virtual and distributed UDDI network by recurring to the ability of self-government of P2P network. Meanwhile, we have a specialized and shared ontology called the Registry Ontology in P2P network which maps each registry to a specific domain. It classified and detailed every UDDI peer according to domain knowledge. Thus, we can go to the UDDI peer more exactly after semantic classification. Another more, a semantically increased UDDI peer is put forward in this paper. It includes UDDI registry, an engine for semantic matching, OWL-S/UDDI mapping module and a communication module to deal with kinds of publishing and requesting information.Secondly, a semantically increased and distributed UDDI layered architecture is proposed in this paper. It can be divided into four different layers, namely the Data Layer, the Communications Layer, the Web Services Application Layer and the Semantic Specification Layer. The Data Layer consists of the Web service Registries. The Communications layer consists of a peer-to-peer network, which provides an infrastructure for the distributed components to communicate with each other. It also provides semantic publication and discovery of web services, the joining and assigning of UDDI peer for Web Service Application Layer. The Web Service Application Layer consists of two actors of system: provider and requestor. The Semantic Specification is to enable the use of semantic metadata through all the above layers.Finally, this paper also investigated the semantic matching methods in the semantic UDDI. By expanding the classical Elastic Matching algorithm and using the concept of the semantic distance, class factor and category factor, we can make the final degree of matching, namely similarity, be in form of numerical value, which can make the result nearer to the request of the requestor.