Domain Ontology Construction And Applied Research In Information Retrieval

As the preliminary work of the research project "Research on the Globus based distributed resoner system",that is supported by Beijing Key Lab of Intelligent Communications Software & Multimedia from BUPT,this paper focuses on the study of the methods for building domain ontologies and its application.With the rapid increment of the information on the web,Information retrieval tools are becoming more and more important. With the help of the information retrieval tools,people can locate the needed information in a quick and correct way. Recall and precision are important standards for information retrieval tools,but the information retrieval based on the keywords is far from satisfaction in recall and precision. On the one hand,a lot of irrelevant links are returned. On the other hand,a large quantity of valuable information is missed. So the information retrieval has to be promoted from the keywords level to the conception level. At the same time the relevance of the conception should be exploited. And the semantic retrieval based on domain ontology is just the key for this improvement.Some achievement the paper has acquired can be expressed as follows:(1) Aimed to the limitations of some existing ontology modeling method,the paper uses the software life cycle modeling method from software_engineering for reference,then builds up the structure model of MovieOntology. A new ontology knowledge storage method based on plane helix relation is presented. This method can reduce the complexity of query constructing,and improve the efficiency of query extension.(2) A three-layers structure for Movie domain ontology based intelligent information retrieval system is designed in the paper,and we realize some fuction modules,such as query building,query optimizing,matching service,orders optimizing of retrival results etc.(3) The paper presents a new query abstracting algorithm,which is based on the mutual mapping between query expression and relation table. Supported by the algorithm,the complicated steps of glossary syncopation can be avoided during the process of natural language query dealing.(4) The author develop an ontology-based information retrieval prototype system of movie domain. Fully utilizing ontology's superiority in knowledge reuse and knowledge share,it can help users acquire professional information from domain knowledgebase.(5) According to the related degree between retrival results and query,the paper put forward a new similarity algorithm that can make the retrieval results ranked descenderly in semantic orders.