SAHAYAK: An Automated Complaint to First Information Report (FIR) Drafting System

Context & Background

First Information Reports (FIRs) are critical legal documents in the Indian criminal justice system. However, drafting an FIR requires translating citizen complaints (written in regional languages with common descriptions) into structured legal jargon, which is currently done manually by police officers, causing long delays.

Problems to be Addressed

Online complaint portals fail because narrative descriptions are often missing essential legal details (e.g. description of weapon in violence cases). This leads to back-and-forth communication delays between citizens and police.

Aims and Objectives

1. Develop SAHAYAK, an AI drafting assistant to process raw complaints.
2. Extract entities, event roles, and identify missing info using NLP.
3. Support English, Hindi, and Hinglish complaints.

Methodology

SAHAYAK operates in two phases: Analysis and Generation. The Analysis phase performs event identification (e.g. distinguishing theft from robbery), entity extraction, and semantic-role labeling. The output graph is matched with predefined templates. If details are missing, the system prompts the user. The Generation phase then constructs a cohesive FIR draft.

Expected Outcomes

A bilingual (Hinglish/English) FIR drafting tool, field testing with state police departments (Delhi Police), and automated slot-filling NLP engines.