Word Sense Disambiguation from English to Indic Language: Approaches and Opportunities

Abstract

Ambiguity is one of the major challenges in Natural Language Processing and the process to solve is known as Word Sense Disambiguation. It is useful to determine the appropriate meaning of polysemy words in a given context using computational methods. Generally, Knowledge, Supervised, and Unsupervised based approaches are the most common methods used to resolve ambiguity problems that occur in a sentence. The government of India has initiated many digital services for its citizen in the last decade. All these services require natural language processing to be easily accessed by web portals or any electronic gadget. Also, these services are provided by the government in Hindi or other Indian languages to better serve Indian citizens. Since English and other languages like Chinese, Japanese, and Korean have plenty of resources available to build applications based on natural language processing but due to low resources available for disambiguating polysemous words in Hindi and other Indian languages, it becomes a hindrance to building any application based on these languages. In this paper, the suggested method enables the assessment of the correct meaning in terms of sustaining data sequences. In order to automatically extract features, the proposed method uses an RNN neural network model. Additionally, it integrates glosses from IndoWordNet. The outcomes demonstrate that the suggested technique performs consistently and significantly better than the alternatives.