Safety Recommendation System By Analyzing Crime Data Using Large Language Model

Abstract

Crime Remains A Critical Concern, Necessitating Advanced Safety Solutions. The Existing Fuzzy-BasedGeo-Spatial Crime Category Prediction System Integrates Fuzzy Logic With GIS For Crime Mapping And Safe Route Suggestions. However, It Faces Limitations Such As Manual Tuning Of Fuzzy Rules, Dependence On Real-time Data Accuracy, And Imprecise Crime Categorization. These Drawbacks Reduce Its Scalability And Adaptability To Evolving Crime Trends. To Overcome These Challenges, This Project, A Safety Recommendation System By Analyzing Crime Data Using Large Language Models, Introduces An AI-driven Approach. By Leveraging Large Language Models (LLMs), The System Dynamically Analyzes Real-time And Historical Crime Data, Eliminating Manual Rule Tuning And Improving Predictive Accuracy. It Collects User-specific Inputs Such As Location Details To Assess Crime Risks And Provide Safety Recommendations. Unlike GIS-based Fuzzy Models, This System Enhances Predictive Precision By Continuously Updating Crime Insights. Real-time Data Analysis Ensures Up-to-date Safety Recommendations, Enabling Users To Make Informed Decisions. By Integrating AI-driven Crime Analysis With User-focused Safety Suggestions, This System Offers A Scalable, Proactive Solution For Crime Prevention.

Keywords