Research Of Ontology-based Semantic Retrieval Model

With the explosive growth of Internet, the information on the Internet increases greatly. People find that it is very difficult to find useful information for themselves in face of mass resources. The traditional information retrieval technologies have limited understanding on semantic level, only provide simple literal match on the user's search keywords and ignore some phenomenon, such as polysemy and synonym. But the semantic retrieval provides a new information retrieval technology based on semantic understanding. The semantic retrieval technology makes use of relationships among information resources, so it has become a hot topic on the research of information retrieval.Ontology, as the basis of semantic web, has the good hierarchical structure of concepts and the support of logical reasoning, and semantic information can be realized through the semantic relationship of concepts. So ontology can be well applied to information retrieval. In the ontology-based semantic retrieval model, the integrity of ontology has a direct effect on the result of information retrieval. So the building of ontology is very important. This paper proposes a new ontology construction method, which is used for constructing the computer domain ontology. Besides that, this paper also proposes a word semantic relevancy computation based on HowNet, which is used for indentifying the domain-special concept in process of constructing ontology.On the basis of constructing the computer domain ontology, this paper proposes a semantic retrieval model based on ontology. This model makes full use of the concepts that formally defined in ontology. The concept is used as index entry, which describes the user's query and document. And it also be used to extend the user's query in ontology. In order to test the performance of the semantic retrieval model, this paper does experiments by comparing this model with traditional model. The result of experiment shows that the semantic model is better than traditional model.The study of this paper is just a beginning step of the research in this field. In the further study, with the development of AI and NLP, the research will focus on how to build ontology with automatic or semi-automatic method, and how to manage the ontology efficiently. And hope that this model can go into the applications.