Design And Implementation Of User Ontology Model Based On Semantic Network Strategy

With the rapid development of the World Wide Web, people are not satisfied by the fact that only shallow information search process is supported by the existing Web. People want to obtain more comprehensive, precise, and timely information from the Web. However, because of the limitations of the existing Web on the efficiency, timeliness and comprehensiveness, it is impossible to achieve such goal within the existing World Wide Web. Therefore, Semantic Web, a next generation of the World Wide Web, is introduced to replace the existing Web so as to satisfy people’s growing information requirements. People have designed and developed a lot of applications to implement and popularize the Semantic Web. In this paper, we discuss one of the hot problems for popularizing the Semantic Web, i.e., how to provide personalized information service in the Semantic Web.Semantic Web is not a completely new and independent network. However, people do not have a standard definition for the Semantic Web at present. A popular one is that Semantic Web is the third generation of the World Wide Web. It aims to enable computers to automatically process information so that people can obtain more precise information to satisfy their requirements.Among the numerous possibilities enabled by the Semantic Web, we are interested in exploiting semantic information in retrieving documents, especially for personalized service. However, to solve the problem, we have to design efficient methods to represent the user’s interests in the working domain at first. After analyzing the existing approaches, we propose a domain ontology based user model, user-onto model, to tackle the problem. Different from the existing approaches that only use the keywords or simply with the taxonomic relations for user modeling, the proposed user-onto model can efficiently utilize both the concepts and semantic relations of the domain ontology to capture user’s interests. Furthermore, we develop a statistical approach to learn the user model from the domain ontology and adopt a spreading activation procedure for inference. This user model with its learning methods has been applied to our developed search engine for finding documents. The experimental results, which are obtained on the Google Directory data sets, support the efficacy of our proposed user model for providing personalized information services in the Semantic Web.