Criminal Code Summarization and Outcome Prediction Using InLegalBERT and Bharatiya Nyaya Sanhita (BNS) Mapping

Abstract

The Bharatiya Nyaya Sanhita (BNS) Act introduces a significant transformation to Indias legal system by replacing the Indian Penal Code (IPC) with a streamlined legal framework. This transition poses challenges for legal professionals who rely on IPC-based analysis tools that lack BNS-specific capabilities for document classification, summarization, and case outcome prediction. To address this gap, this project presents a BNS-compatible legal analysis system that incorporates a rule-based mapping between IPC and BNS sections, supported by a synthetic dataset for modeling BNS-related cases. Document summarization is performed by fine-tuning InLegalBERT, a language model designed for legal contexts, enabling the precise extraction of critical information from legal documents. In addition, a binary classification model is used to predict case outcomes, supporting efficient analysis and decision-making for BNS cases. By integrating InLegalBERT-based summarization, structured IPC-to-BNS mapping, and outcome prediction, the proposed system assists legal professionals in transitioning to the BNS framework. The project also provides a foundation for further advancements in legal technology aligned with the evolving requirements of Indias modern legal system.