Research of information retrieval method based on fuzzy rough sets theory

This paper proposes a new information retrieval method based on fuzzy rough sets theory. In the approach, firstly, the traditional Mutual Information function are used to compute the semantic association weight among document characteristic words, i.e., construct the thesaurus, the similar concept class of each characteristic word is mined based on it. Secondly, the fuzzy-rough set of document and query is built with thesaurus and similar concept class, its lower and upper approximation expand the expression of document and query, it implements conceptual expansion by association semantics. Finally, the semantic closeness between document and query is computed by the semantic distance on fuzzy-rough set, the ordinal results are returned to user according to the close degree value. The user can also adjust support threshold to get his satisfactory research objects through feedback information. This method can realize conceptual retrieval. At the end of the paper, an example is given for further illustration.