AI analysis of Dial 112, FIRs to generate crime hotspots and early warnings for L&O

AI analysis of Dial 112, FIRs to generate crime hotspots and early warnings for L&O

Objective

The initiative aims to transform unstructured Dial‑100 emergency call data into structured, actionable intelligence that empowers district-level leadership—such as SPs and Commandants—to detect and respond to emerging public safety concerns in near real-time. By automating the ingestion, classification, and analysis of call records through NLP and large language models (LLMs), the system delivers location-specific insights, highlights recurring issue types, and identifies geographic hotspots of criminal or civil unrest activity. The solution bridges operational gaps by eliminating manual review bottlenecks, accelerating situational awareness, and supporting evidence-based decision-making at both police station and battalion levels.

Scope

The system will automatically process incoming Dial‑100 logs every 10 minutes, extracting relevant textual information and associating it with jurisdictional metadata. It will apply AI-based classification to determine issue categories (e.g., domestic violence, theft, land dispute) and extract named entities like persons or landmarks. The processed data will feed into a dynamic dashboard featuring real-time heatmaps, severity scoring, and recurrence analysis to highlight top hotspots across the district. Additional outputs include daily and weekly summaries for senior officers. A role-based review interface allows Power Users to correct low-confidence classifications, enabling post-event model retraining. Robust error handling, retry mechanisms, and flagging of unprocessable records are integrated to ensure resilience. Features such as full jurisdictional tagging, notification systems, and summary report generation are included, while live voice analysis and predictive policing components remain out of scope for the current version.