A framework for combining open source intelligence and natural language processing with ensuring the judicial admissibility of digital evidence in legal and law enforcement applications.

A framework for combining open source intelligence and natural language processing with ensuring the judicial admissibility of digital evidence in legal and law enforcement applications.

Mahdi Korehpaz1 Reza Sheibani2 Seyed Mohammad Hossein Moattar3

1) Department of Computer Engineering, Mashhad Branch, Islamic Azad University, Mashhad, Iran
2) Department of Computer Engineering, Mashhad Branch, Islamic Azad University, Mashhad, Iran
3) Department of Computer Engineering, Mashhad Branch, Islamic Azad University, Mashhad, Iran

IOI Code : AAHD-ABKKZZ
Publication : National Conference on Information Technology, Nanotechnology, Artificial Intelligence and Technological Futures Studies(itnaf.ir)
Abstract :
    Open source intelligence (OSINT), as one of the main sources of information in legal and law enforcement investigations, faces the fundamental challenge of judicial admissibility. Despite the fact that more than 80% of law enforcement intelligence activities today are based on OSINT, the lack of a standard methodology for collecting and analyzing open source data, coupled with the lack of a framework for ensuring the authenticity and chain of custody of digital evidence, has seriously questioned the use of this data in judicial authorities. The present study aims to present a three-layer architectural framework for combining artificial intelligence and natural language processing in the intelligent collection and analysis of open source data. The first layer is based on local and secure deployment of systems with a privacy-preserving approach, the second layer is based on a legal-regulatory knowledge graph to extract semantic relationships between entities, and the third layer ensures the chain of custody and explainability of digital evidence. Research findings show that using knowledge graphs instead of purely textual language models increases the accuracy of legal analysis by 85% and, with additional metrics including ALIGN-SCORE (semantic alignment with human judicial reasoning) and Contestability Index (contestability), provides the ability to defend evidence in court.

Keywords : open source intelligence natural language processing admissible digital evidence legal knowledge graph law enforcement applications