Research On Semantic Retrieval Based On Domain Ontology

With the rapid development of Internet technology and increase in network information, how to search the required resource from Internet quickly and accurately has becoming an important topic in current information retrieval. Traditional information retrieval systems are searched based on key words match or top taxonomy, which will always return a lot of irrelevant results and can not be satisfy the users' demand of recall and precision.With the ontology being used widely, people attempt to use the relations among concepts of ontology to improve the retrieval system's semantic. Ontology can not only describe the information system in hierarchical of semantic and knowledge, but also has a good concept hierarchical structure, and support reasoning. After the traditional information retrieval system combined with ontology, the retrieval technology also transfer from based on key words to the new version based on knowledge.The paper researches the information retrieval systems based on domain ontology. First, the background,significance and the current research situation both abroad and at home are introduced.Besides, the relevant theories of ontology and information retrieval are outlined. Then the two key technologies, which are the semantic annotation of document and query expansion, are deeply studied. The paper proposes an improved semantic annotation method based on domain ontology in the part of semantic annotation. The method annotates document by the semantic context of the ontology entity and the structural information of the document. In the part of query expansion, it mainly use the ontology's concepts and attributes to deal with the user question, so as to realize semantic query expansion. After that,based on the above theoretical research, we chose 100 texts of the dress as the experimental data. I designed and basically achieved the dress domain ontology-based semantic retrieval system. The testing results proved that it has much higher query recall rate and precision rate. Finally, the whole text is summarized and the prospects for the future search work is advanced.