Natural Language Understanding (NLU)- Powered Information Retrieval from Indian Legal Property Graph

Existing solutions in legal document mining are largely unsuitable for the Indian legal system, leading to inaccuracies and numerous challenges. Indian case files are verbose and contain domain-specific terminology, rendering them difficult for general public to understand. This research seeks to address existing gap by developing a comprehensive search query engine that facilitates access to legal cases. It will allow users to easily retrieve legal information using preferred keywords, including year, precedent, and judge, through implementation of an Indian legal information retrieval system. The proposed retrieval system extracts detailed information regarding legal components, including date of judgement, relevant statutes, judges involved, and additional elements, by utilising fundamental capabilities of entity extraction and intent detection in Natural Language Understanding (NLU). The retrieval system is engineered to extract information from a new Indian Legal Property Graph. The model demonstrates superior performance compared to current state-of-the-art techniques, having been trained on Indian legal data.