Information Retrieval Using Rough Sets

概要

論文の詳細を見る
All previous work on intelligent information retrieval using rough sets is based on the equivalence rough set model(ERSM) that organise the vocabulary (index terms) into an approximation space of equivalence classes. In this paper we first show that the vocabulary cannot be suitably partitioned into equivalence classes and this task cannot be computationally efficient. We then propose a tolerance rough set model (TRSM) for information retrieval that organizes the vocabulary and searches documents in an approximation space of tolerance classes. The core of TRSM is tolerance classes constructed by the index term co-occurrence and a matching algorithm with tolerance rough inclusions. We report the implementation and an evaluation of TRSM on the database of articles and papers of the Journal of the Japanese Society for Artificial Intelligence in its first ten years of publication. We analyse good properties of TRSM and show that TRSM can overcome the limitations of ERSM in information retrieval.
社団法人人工知能学会の論文
1998-05-01