Enhancing Marathi Abstractive Text Summarization with IndicBART

The automatic text summarization technology is used in different industries such as news, law, medicine, education, social media, and business intelligence for their respective information extraction needs. Text summarization converts extended textual materials into shortened condensed content which maintains crucial points. It automatically produces summaries which retain essential sentences as well as all important facts found in the original document. The Text summarization field primarily focuses on English content but demonstrates limited research in Marathi language. Because of rich morphology and intricate syntactic patterns, Marathi language poses special difficulties for Natural Language Processing (NLP). Using pretrained transformer-based models, this study investigates Abstractive Text Summarization for Marathi with an emphasis on IndicBART, a language model created especially for Indian languages. Using a Marathi dataset, the IndicBART model has been refined to provide excellent abstractive summaries by utilizing its transfer learning and contextual awareness. The proposed approach is evaluated using ROUGE-1, ROUGE-2, and ROUGE-L metrics to measure its effectiveness