Information Retrieval Based on Word Senses

This paper proposes an algorithm for word sense disambiguation based on a vector representation of word similarity derived from lexical co-occurrence. It diiers from standard approaches by allowing for as ne grained distinctions as is warranted by the information at hand, rather than supposing a xed number of senses per word, and by allowing for more than one sense to be assigned to a given word occurrence. The algorithm is applied to the standard vector-space information retrieval model and an evaluation is performed over the Category B TREC-1 corpus (WSJ subcollection). Results show that this sense disambiguation algorithm improves performance by between 7% and 14% on average .