AILSE - AI driven Legal Search Engine for Indian Supreme Court Cases

The use of heterogeneous databases, key word search engines, and access to large volumes of cases is often impeding legal research in the Indian jurisdiction. This paper presents AILSE an end-to-end pipeline performing legal search, question-answering and summarization tasks employing state-of-the-art natural language processing (NLP) technology that is specifically tuned to the Indian legal system. AILSE allows all users to query judgments of the Indian Supreme Court using plain English and follows a methodology that includes web scraping, pre-processing and advanced NLP tasks like summarization and question-answering. It takes advantage of embedding models and vector-based databases to find semantically relevant cases and includes the functionality of automated case summarization, structured metadata extraction, and interactive chat-based navigation of specific cases with the help of retrieval-augmented generation (RAG). AILSE includes user authentication, personalized search-history feature, and an interactive interface built on React and Firebase. Experiments were carried on 924 cases scrapped from the Supreme court of India website for the year 2024. Empirical studies reveal that the system significantly increases the efficiency, accessibility and completeness of legal research compared to the existing commercial and open-access products. This work has introduced a new standard in the retrieval of legal information in India by facilitating access to legal information in an interactive and context-sensitive manner, and has facilitated the examination of case law in India.