Reserch On Domain Ontology Concept And Relation Acquisition

Ontology which can describe the system conceptual model in the semantic and knowledge level, more and more widely applied in the area of information retrieval, artificial intelligence and knowledge management. Owing to ontology construction is a time-consuming work, the method of automatic or semi-automatic constructing become a hot spot of ontology research. This article focuses on the two mainly questions that of domain ontology Automatic construction:domain concept extraction and relationship between concepts, the major work progress as follows :1)Research the related ontology theory, study the method of ontology construction, design a domain ontology semi-automatic construct frame based on free text, mainly includes: combination word extraction module, domain concept word extraction module, relationship between concept extraction module.2)Extract combination words using the method of Mutual Information and Context Dependency Analysis, domain concept words usually appears as the form of combination words which is the first step of construct ontology, using mutual information and Context Dependency Analysis method to extracting the combination word. Domain concept words must be relevant to the domain, and less emerged in other areas, through compute the Domain Relevance ,Domain Consensus and filter the combination words which below the threshold to extract concept words.3)Realize the taxonomy relationship gaining between concept, achieving the non- taxonomy Inherit relationship gaining to calculate the similarity degree for domain concept through hierarchical clustering method, Realizing the gaining of inheriting relationship classification by generalized suffix trees .Put forward the non- taxonomy relationship gaining method based on association rules, calculating association degree between concepts, setting threshold to extract the non-taxonomy relationship of concepts automatically, labeling relationship for the conceptual by domain experts.This thesis prove the feasibility of the above algorithm through experiments, and finally will achieve the domain ontology semi-automatic frame , this thesis get the support of Natural Science Foundation Research Programs of Shaanxi province, will eventually applied to the construction of mechanical product knowledge base which based on ontology.