Concept-Based Indexing in Text Information Retrieval

Traditional information retrieval systems rely on keywords to indexdocuments and queries.Insuchsystems,documents are retrieved based onthe number ofshared keywordswith the query.This lexical-focused retrieval leads to inaccurate and incomplete results when different keywords are used to describethedocuments and queries. Semantic-focusedretrieval approaches attempt to overcome this problem byrelying on concepts rather than on keywordstoindexing and retrieval.The goal is to retrievedocumentsthat aresemantically relevant to a given user query.This paper addresses this issue byproposing asolution at the indexing level. More precisely, we propose a novel approach forsemantic indexing based onconceptsidentifiedfrom alinguistic resource.In particular, our approach relies on the joint use ofWordNet and WordNetDomains lexical databases for concept identification.Furthermore, we propose asemantic-based concept weighting scheme that relies on a novel definition of concept centrality. Theresulting system is evaluated on the TIME test collection. Experimental results show the effectiveness of ourproposition over traditional IR approaches.