Abstract: Information retrieval based on ontology is an effective approach to achieve knowledge retrieval. A kind of information retrieval methodology based on rough ontology is proposed against formal concept, which is supported by ontology, is not enough to represent incomplete knowledge. This methodology represents ontology in the form of ontology information system. First, given a keyword based query, it combines with the keyword based method to search documents in predefined semantic document space, and then, association search method is used to search individual set and property set relating to query keywords, and property set as equivalence class constructs approximation space. Last, approximation space is used to compute similarity of individual set and document set and according to the degree of similarity to rank. The experiment shows that this methodology can be performed more better than methods based on keyword and classical ontology in precision.

Key words: rough ontology, information retrieval, association search, similarity