Domain Driven Disambiguation
Our current investigation in Word Sense Disambiguation focuses on the role of domain information. Semantic Domains provide a natural way to establish semantic relations among word senses, which can be profitably used for Computational Linguistics.
The Domain Driven Disambiguation approach to WSD decomposes the task into two steps. For each word to disambiguate the domain of the context is estimated. Then it is compared to the domain of each word sense, and the most similar one is selected.
We developed WordNet Domains, an extension of the Princeton English Wordnet, where synsets have been annotated by domain labels, such as Medicine, Architecture and Sport. The resource is freely available for research purposes.
We developed the DDD system for WSD. DDD has been evaluated in the SENSEVAL-2 and SENSEVAL-3 tasks for English, Italian, Spanish, Catalan and Basque.