Relating WordNet senses for word sense disambiguation

The granularity of word senses in current general purpose sense inventories is often too �ne-grained, with narrow sense distinctions that are irrelevant for many NLP applications. This has particularly been a problem with WordNet which is widely used for word sense disambiguation (WSD). There have been several attempts to group WordNet senses given a number of different information sources in order to reduce granularity. We propose relating senses as a matter of degree to permit a softer notion of relationships between senses compared to �xed groupings so that granularity can be varied according to the needs of the application. We compare two such approaches with a gold-standard produced by humans for this work. We also contrast this goldstandard and another used in previous research with the automatic methods for relating senses for use with back-off methods for WSD.