New hybrid measure for the similarity between two queries in an information retrieval system

Abstract

Find relevant documents for a given query in search process information seems a challenge for all models and techniques embodied in an information retrieval system. We continue this challenge by proposing a new measure of similarity between two queries. This similarity measure is based on two aspects: the first is statistical based on the statistic X2 (method CHIR) and the second is semantic based on the mutual information. Our proposed method shows its efficiency and performance according to tests performed on the standard corpus of research information CACM.