ENHANCING THE FUZZY SET MODEL FOR HIGH-QUALITY DOCUMENT RANKINGS

Abstract

It has been argued for a long time that the conventional fuzzy set model based on the MIN and MAX operators is not appropriate for information retrieval systems. This is because the ranked output of documents from the conventional fuzzy set model does not agree with human's intuition in many cases. In this paper we first analyze various fuzzy operators, i.e., T-operators and averaging operators, and describe that the problems of the conventional fuzzy set model cannot be overcome with any type of T-operators. We then present a new formulation of the fuzzy set model which employs averaging operators. We show through operator graph analyses that the proposed fuzzy set model enhances the quality of document rankings. It is also described that the behaviour of the proposed fuzzy set model almost coincides with those of the extended boolean model.