A Knowledge-Base Oriented Approach for Automatic Keyword Extraction

Authors

  • Ludovic Jean-Louis École Polytechnique de Montréal
  • Michel Gagnon École Polytechnique de Montréal
  • Eric Charton École Polytechnique de Montréal, and Centre de Recherche Informatique de Montréal

DOI:

https://doi.org/10.13053/cys-17-2-1523

Keywords:

Automatic keyword extraction, encyclopedic knowledge.

Abstract

Automatic keyword extraction is an importantsubfield of information extraction process. It is adifficult task, where numerous different techniques andresources have been proposed. In this paper, wepropose a generic approach to extract keyword fromdocuments using encyclopedic knowledge. Our two-stepapproach first relies on a classification step for identifyingcandidate keywords followed by a learning-to-rankmethod depending on a user-defined keyword profile toorder the candidates. The novelty of our approach relieson i) the usage of the keyword profile ii) generic featuresderived from Wikipedia categories and not necessarilyrelated to the document content. We evaluate oursystem on keyword datasets and corpora from standardevaluation campaign and show that our system improvesthe global process of keyword extraction.

Downloads

Published

2013-06-29