Paper: Disambiguating Noun Compounds With Latent Semantic Indexing

ACL ID W02-1401
Title Disambiguating Noun Compounds With Latent Semantic Indexing
Venue CompuTerm International Workshop On Computational Terminology
Year 2002

Technical terms in text often appear as noun compounds, a frequently occurring yet highly ambiguous construction whose interpretation relies on extra-syntactic information. Several statistical methods for disambiguating com- pounds have been reported in the literature, of- ten with quite impressive results. However, a striking feature of all these approaches is that they rely on the existence of previously seen un- ambiguous compounds, meaning they are prone to the problem of sparse data. This difficulty has been overcome somewhat through the use of hand-crafted knowledge resources to collect statistics on “concepts” rather than noun to- kens, but domain-independence has been sacri- ficed by doing so. We report here on work inves- tigating the application of Latent Semantic In- dexing ...