Automatic Adaptation of WordNet to Domains
Roberto Navigli, Paola Velardi · 2002
The objective of this paper is to present a method to automatically enrich WordNet with sub-trees of concepts in a given language domain. WordNet is then trimmed to reduce unnecessary ambiguity and singleton nodes. The process is based on the use of statistical method and linguistic processing to extract candidate domain terms. Multiword terms are semantically disambiguated and interpreted using ontological and contextual knowledge stored in WordNet on singleton words. 1.