A Graph-Based Method to Improve WordNet Domains

Aitor Gonzalez-Agirre, German Rigau, Mauro Castillo

Abstract: WordNet Domains (WND) is a lexical resource where synsets have been semi-automatically annotated with one or more domain labels from a set of 170 hierarchically organized domains. The uses of WND include the power to reduce the polysemy degree of the words, grouping those senses that belong to the same domain. This paper presents a novel automatic method to propagate domain information through WordNet. We compare both labellings (the original and the new one) allowing us to detect anomalies in the original WND labels. We also compare the quality of both resources (the original labelling and the new one) in a common Word Sense Disambiguation task. The results show that the new labelling clearly outperform the original one by a large margin.