Automated Word Sense Disambiguation Using WordNet Ontology
Khaoula Belila, Okba Kazar, Mohammed Charaf Eddine Meftah · International Journal of Organizational and Collective Intelligence · 2022
Automatic word sense disambiguation is a major challenge in natural language processing domain. In recent years, many of supervised and knowledge-based approaches were developed to solve this problem. The use of a sense inventory as a knowledge background to disambiguate words is a very useful technique rather than supervised approaches, with its need of a large pre-trained text corpus. This paper proposes a new approach to disambiguate words in text based on WordNet ontology, as sense inventory. The authors introduce a new technique called Gloss+ for word-sense disambiguation (WSD), which is based on using of the glosses of WordNet synonyms of the target word and the local context in which this word is used. This technique exploits a special behavior of the polysemous synsets. This behavior is detected during the disambiguation process and is used to improve the results obtained. In the experiment part, the authors compare the proposed approach to the methodologies which use synonyms or glosses only.