Implementation of Neighboring Word Feature Technique using CNN for Word Sense Disambiguation for Marathi Language

Abstract

Basically, WSD is a linguistic method for automatically defining the true sense of a word in the particular sense of context. WSD analysis is worldwide challenge in AI-computational field and NLP. The various operations of WSD system include Information retrieval system (IRS), Lexicography, Speech & Text Processing, Machine Translation.  In every language of this world a single word may have different meaning in various context of sentence. For the human being it is very easy and simple to determining the exact meaning after reading and analyzing the complete sentence. Our research work includes, identifying the appropriate sense of lexeme for already available context where it has been used and finding exact relationship with the lexicons which can be done using NLP techniques This paper attempt to develop and implement a unique technique for WSD of different words in Marathi language, which is one of the most popular regional Indian languages. In the experiment, we have developed and implement our novel knowledge based techniques with wordnet of Marathi language. Our proposed approaches uses knowledge based word sense ambiguity algorithm for WSD in Marathi language. The proposed  unique approach  is capable of identify and solve the ambiguity and provides sense of ambiguity is very less amount of time. Accuracy achieved for proposed approach is 87.22%.