Enhancing Public Access to Legal Knowledge in India: A Legal Chatbot Using Legal BERT, GPT-2, and Retrieval-Augmented Generation (RAG)

This work describes a legal chatbot that would be able to support and aid in answering legal queries on Indian law through the use of Legal BERT, GPT-2, and RetrievalAugmented Generation. This is a chatbot that has been trained by fine-tuning it using a curated corpus consisting of the Indian Constitution and other relevant legal texts, enabling the chatbot to deliver the most accurate, contextually appropriate response. Legal BERT enhances the chatbot’s ability to understand sophisticated legal terminology, while GPT-2 generates responses similar to that of a human with material retrieved from those sources. RAG is incorporated to enhance the retrieval of relevant sources so that accurate and relevant answers are delivered. Accuracy as well as efficiency is displayed by the system; it has very high potential towards enhancing access to public legal information, supporting legal research, and assisting in access to justice initiatives under the Indian legal structure.