Towards a semantic representation of documents by ontology-document mapping

Abstract

This paper deals with the use of ontologies in Information Retrieval field. It introduces an approach for document content representation by ontology-document matching. The approach consists in concepts (mono and multiword) detection from a document via a general purpose ontology, namely WordNet. Two enterions are then used: co-occurrence for identifying important concepts in a document, and semantic similarity to compute semantic relatedness between these concepts and then to disambiguate them. The result is a set of scored concepts-senses (nodes) with weighted links called semantic core of document which best represents the semantic content of the document. We regard the proposed and evaluated approach as a short but strong step toward the long term goal of Intelligent Indexing and Semantic Retrieval. © Springer-Verlag Berlin Heidelberg 2004.