The Automatic Acquisition of a Broad-Coverage Semantic Lexicon for Use in Information Retrieval

Representation and Acquisition of Lexical Knowledge: Polysemy, Ambiguity, and Generativity

Volume

Representation and Acquisition of Lexical Knowledge: Polysemy, Ambiguity, and Generativity

Contents

Our interest is in developing techniques for constructing a semantic lexicon of broad coverage from machine tractable resources. We work within a paradigm in which each word sense is represented as a vector in an ndimensional feature space. So far our experiments have encompassed the Merriam- Webster Compact Electronic Dictionary, the Irish An Focl6ir Beag and the Princeton WordNet. Our main uses for the results are in full text information retrieval, machine assisted translation and lexical alignment. In this paper we explain the background to the approach, outline the algorithms with which we have been experimenting and report on the results which we have obtained.