With the emergence of the semantic web, ontology has attracted a great deal of attention in field of information retrieval. But the conceptual formalism supported by typical ontology is not sufficient for handling incomplete information that is confronted in the real world knowledge. To tackle this problem, a semantic information retrieval approach based on a rough ontology is proposed. Rough ontology in this paper is in the form of an ontology information system. Given a keyword based query, our approach infers the individuals and properties correlated to the query through a procedure of association searches in the rough ontology, and takes properties as equivalence relations to construct an approximation space of rough ontology. Afterward, an algorithm of computing similarity in rough ontology is presented, and approximation space is employed to compute similarity for ranking documents in semantic document indexing space. The proposed approach has been compared with two other information retrieval techniques, and the experiments conducted on CNKI collections, support the better efficacy which results from our approach.

No abstract

No abstract

Abstract