Lucia Specia, Mark Stevenson, Maria das Graças Volpe Nunes
To appear in Journal of Language Resources and Evaluation, Kluwer
Corpus-based techniques have proved to be very bene cial in the development of efficient and accurate approaches to word sense disambiguation (WSD) despite the fact that they generally represent relatively shallow knowledge. However,
it has always been thought that WSD could also bene t from deeper knowledge
sources. We describe a novel approach to WSD using inductive logic programming
to learn theories from first-order logic representations that allows corpus-based evidence to be combined with any kind of background knowledge. This approach hasbeen shown to be effective over several disambiguation tasks using a combination of deep and shallow knowledge sources. Is it important to understand the contribution of the various knowledge source used in such a system. This paper investigates the contribution of ten knowledge sources to the performance of the disambiguation models produced for the SemEval-2007 English lexical sample task. The outcome of this analysis will assist future work on WSD in concentrating on the most useful knowledge sources.
Report number: