Skip to Main content Skip to Navigation
Theses

Corrélation sémantique entre documents : application à la recherche d'information juridique sur le Web

Abstract : There are many ways to find information on the Web and search engines are the most frequently used tools. In this context, relative pages algorithms are complementary techniques providing more information about on specific document without asking any question. The goal of our work is to define a new semantic relative page algorithm to perform search on a law oriented corpus. To reach the goal, we defined a method that applies linguistic tools and techniques on previously selected documents. Relevant text units are extracted from our documents'corpus and are called lexical signatures. We use those lexical signatures as requests to search engine; the results correspond to the pool of relative pages. Our relative pages algorithm is used and evaluated in an information retrieval context, being included in the development of a search engine. The main contribution of our work are (1) a new perspective for building lexical signatures to perform relative pages searches, (2) the definition and evaluation of a new relative pages algorithm calles Tifr, (3) a discussion on the semantic aspect of our method and finally, (4) a practical answer to the challenge of information retrieval in a law oriented context.
Document type :
Theses
Complete list of metadatas

https://pastel.archives-ouvertes.fr/pastel-00001080
Contributor : Ecole Mines Paristech <>
Submitted on : Friday, August 6, 2010 - 4:02:12 PM
Last modification on : Thursday, September 24, 2020 - 4:36:01 PM
Long-term archiving on: : Monday, November 8, 2010 - 3:13:58 PM

Identifiers

  • HAL Id : pastel-00001080, version 1

Citation

Christophe Chotteau. Corrélation sémantique entre documents : application à la recherche d'information juridique sur le Web. Informatique et langage [cs.CL]. École Nationale Supérieure des Mines de Paris, 2003. Français. ⟨NNT : 2003ENMP1185⟩. ⟨pastel-00001080⟩

Share

Metrics

Record views

799

Files downloads

278