Enhancing crime record analysis: information extraction and categorisation using a fuzzy logic approach

Efficiently extracting and categorising information from crime records is crucial for actionable insights in law enforcement. Traditional methods struggle with language uncertainty. We propose a fuzzy logic-based approach for information extraction and categorisation from criminal event documents. Fuzzy rules enhance imprecise boundary delineation among patterns. Fuzzy crime extracts crime-related named entities (NER) like incident date, weapon type, location, nationality, and involved persons. It builds a crime-related thesaurus using computational linguistic methods. The ANFIS model categorises sentence patterns, using fuzzy rules designed with four variables to generate 16 patterns. Higher weighted patterns indicate more significant sentences. The system effectively extracts specific crime-related details from reports; classifying sentences using ANN. Experiments on the Iraq Body Count (IBC) benchmark dataset validate our model's accuracy using precision, and recall measures, outperforming previous techniques. Our fuzzy logic-based approach enhances information extraction and categorisation in crime records, enabling law enforcement agencies to make informed decisions.