FORENSIC INTELLIGENCE DERIVED FROM CRIME SCENE EVIDENCE USING TEXT EMBEDDINGS

Forensic Intelligence is an emerging paradigm that shifts the focus of forensic science from isolated case-by-case analysis to broader, macro-level examination aimed at generating actionable intelligence. Although promising, this approach remains underexplored in practice, with limited studies demonstrating how to operationalize Forensic Intelligence or extract value from the collective analysis of crime scene evidence, for example. This work introduces an innovative tool that leverages Artificial Intelligenc, particularly Natural Language Processing (NLP) recent techniques to uncover patterns in crime scene reports. One of the key outcomes of the methodology described here is its ability to support crime linkage analysis. While crime linkage is often treated as a separate domain in the literature, this study highlights its strong alignment with the principles of Forensic Intelligence. It demonstrates how AI can assist investigators in identifying relationships between crimes, such as shared modus operandi. This research further contributes by applying modern NLP technologies, specifically transformer-based models and large language models (LLMs), to analyze violent crime data. Through methods of summarization, text embeddings, clustering, and case similarity measurements, the study offers a practical tool for visualizing and interpreting forensic data to enhance both Forensic Intelligence and Crime Linkage insights.