Modified Lesk approach for Word Sense Disambiguation in the Marathi Language
Rasika Ransing, Archana N. Gulati, Ritu Srivastava · 2023
The process of computationally determining the meaning of a word in a certain context is known as Word Sense Disambiguation (WSD). One of the most challenging tasks in Natural Language Processing is WSD. In terms of tools and datasets, Marathi is a low-resource language. As a result, there hasn’t been much research undertaken on the Marathi language. One of the most popular methods of the knowledge-based approach for WSD is the Lesk algorithm, which uses WordNet instead of the conventional dictionary as the sense inventory. In this paper, we present a modified Lesk algorithm that considers all the words in the given sentence except the ambiguous word as well as the sense, synonyms, and examples of the words to the left and right of the target ambiguous word as the context for each sense of the ambiguous word. This makes it possible to ascertain the correct meaning of ambiguous words given the context. Nouns, verbs, adjectives, and adverbs can be disambiguated using the proposed method.