AN APPROACH FOR DOMAIN ONTOLOGY-BASED SEMANTIC RETRIEVAL

The retrieval of web data is usually difficult, because web data do not have a standard format to store. The traditional information retrieval on a purely syntactic base failed to satisfy the high quality retrieval requirement. Semantic retrieval was raised in this context. Semantic retrieval offers the potential to express web queries and analyze documents in a semantic way. However, the development of semantic retrieval is still at a relatively early stage. And most search engines are still on a syntactic base or on a semantic base but with limited functions and low precision and recall. This research proposes a fresh approach meant to semantically query web documents using a natural language analysis technique and domain specific ontology. This approach refines internet users' queries by synonym expanding and limiting queries' semantic boundary using domain ontology extended from HowNet. HowNet is an ontology which supports both Chinese and English. Based on HowNet, a semi-automatically built method for domain ontology is raised.