This paper presents an Artificial Intelligence–driven Smart Crime Investigation Support System designed to assist law enforcement agencies in accelerating and improving criminal investigations. Traditional investigative methods struggle to efficiently process large volumes of heterogeneous evidence such as textual reports, digital records, multimedia files, and forensic documents. The proposed system integrates large language models, retrieval-augmented generation, and multimodal data processing to analyze evidence, generate intelligent investigative hypotheses, identify similarities with historical cases, and visualize relationships among suspects, evidence, and locations. Experimental evaluation demonstrates a significant improvement in investigative efficiency, achieving up to a 90% reduction in analysis time and hypothesis accuracy ranging between 92% and 96% compared to conventional approaches. The framework incorporates strong security, privacy, and governance mechanisms, including encryption, role-based access control, and explainable AI principles. The system is designed for future interoperability with national crime databases such as CCTNS, NCRB, and ICJS, enabling cross-agency collaboration.
Samar Quadri (Mon,) studied this question.