Title: The impact of titles expansion based on ontology in document retrieval

Authors: Belkacem Abdelli; Okba Kazar; Jean-Marie Pinon

Addresses: Department of Computer Science, University of Biskra, Biskra, Algeria ' Department of Computer Science, University of Biskra, Biskra, Algeria ' LIRIS Laboratory, INSA Lyon, Villeurbanne, France

Abstract: Among the features of documents is their logical structure, which represents their components such as chapters, sections, paragraphs, titles, chapter titles, etc. The titles and subtitles of documents are meaningful; they are good indicators of the paragraphs' content. For this reason particular attention should be paid to these titles during the indexing process and research. The terms of the titles are the most important in the document, but their number is very limited, because of their shortness, which leads to irrelevant results from information retrieval (IR). One possible solution is to extend titles by adding other terms that have a semantic similarity with initial terms. The present work is an attempt to study the effect of extending the most important terms in the documents on information retrieval. Experiments on a large corpus, INEX 2009, show the effectiveness of the proposition and an improvement in the precision of the results in IR.

Keywords: information retrieval; logical structure; metadata; semantics; WordNet similarity; title expansion; ontology; document retrieval; indexing; document titles; document subtitles.

DOI: 10.1504/IJMSO.2015.073875

International Journal of Metadata, Semantics and Ontologies, 2015 Vol.10 No.3, pp.170 - 181

Received: 21 Nov 2014
Accepted: 19 Apr 2015
Published online: 27 Dec 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article