Improvement of a retrieval and f iltering systems by an automatic multi words extraction tool

Abstract

The role of all retrieval systems consists in finding documents (or document excerpts) that best answer an information need. This need, whether precise and instant (a request) or general and long term (a profile), is most often expre ssed by a list of entities called terms. The goal of this article is to present an automatic mu lti words recognition and extraction tool, compiled from text extracts, in order to build a thesaurus, or terminological basis, useful by search engines and filtering systems in order to boost their performances.