Abstractive Legal Summarization for Low Resource Bodo Language Using Pretrained Models

This paper explores legal frameworks and natural language processing (NLP) in the context of Indian low resource languages, with a particular focus on Bodo legal texts. It examines the challenges of legal text summarization and the development of domain-specific NLP models tailored to the Bodo language. The study highlights the limitations of existing datasets and proposes methodologies for improving legal text processing through state of art summarization techniques AI4Bharat’s IndicBart and Face-book’s Bart model. Additionally, it discusses model evaluation strategies and fine-tuning approaches to enhance the accuracy and usability of automated legal summarization systems for low resource language. The findings underscore the importance of linguistic diversity in legal AI applications and the need for robust NLP tools to support underrepresented languages in the legal domain.