Enhancing NLP for Indic Languages With Limited Resources: A Study of Transformer Models for Translation and Summarization

Significant advancements have been made in natural language processing (NLP), especially in numerous languages. The vast linguistic diversity of India presents unique challenges for automated language processing tasks as Indian languages other than Hindi are marginalized in both cases; datasets which incur a vast toll on summarization models that require huge datasets to be trained on. This paper presents a comparative and comprehensive analysis of multiple transformer-based models which have been trained and/or fine-tuned for lesser-used Indic languages. Purely focusing on Natural Language Translation (NLT) and evaluating its efficiency in generating contextually accurate summaries for languages such as Bengali and Marathi. Models such as mT5, IndicBART, mBART and other multilingual NLP models will be benchmarked and assessed using the ROUGE score. This study highlights the model's strengths and weaknesses in handling the limitations summarization of low-resource Indic languages. The findings will contribute heavily to the development of NLP tools aimed at improving summarization inclusivity across India's diverse languages.