Police and Security, Vol. 34 No. 3, 2025.
Preliminary communication
https://doi.org/10.59245/ps.34.3.4
Cognitive-Intelligence Model for Predicting Security Threats
Tonći Prodan
; University of Split, University Department for Forensic Sciences, Croatia.
Marija Gombar
orcid.org/0009-0000-8621-4007
; Janko Bobetko Centre for Defence and Strategic Studies, Dr. Franjo Tuđman Croatian Defence Academy, Croatia.
Abstract
Contemporary security challenges require sophisticated methods for forecasting criminal activities, as traditional criminal analytics are predominantly retrospective and limited in their anticipatory capacity. This paper presents the Cognitive-Intelligence Model for Security Threat Prediction (KOMP), developed as a conceptual framework that integrates artificial intelligence, analysis of security-relevant information, and criminal analytics to enhance risk assessment and enable proactive action. The terms “cognitive” and “intelligence” in the model's name refer to its capacity for processing and interpreting multilayered security information. The model combines existing methodologies with new interpretative approaches, whereby its contribution lies not in individual technical components but in how they are conceptually integrated. This research aims to design and evaluate a model that increases predictive accuracy, reduces algorithmic bias, and optimises the use of security resources. KOMP uses neural networks, machine learning, and big data analytics to identify criminal behaviour patterns and detect anomalies in financial, communication, and geolocation data. The methodological framework includes simulation experiments, analysis of synthetic criminal data, and comparative evaluation against existing forecasting models (PredPol, CompStat). Results indicate that KOMP achieves higher precision in threat detection and significantly reduces the number of false positives. The model enables real-time adaptation of security strategies, though its implementation requires addressing ethical and legal concerns, particularly regarding data privacy and algorithmic transparency. This research contributes to predictive criminal analytics by offering a methodologically grounded framework for developing intelligent security systems, with potential applications in crime prevention, operational efficiency enhancement, and strategic planning against organised crime, terrorism, and cyber threats.
Keywords
predictive algorithms; conceptual model; criminal analytics; KOMP model; security threats; artificial intelligence.
Hrčak ID:
334201
URI
Publication date:
23.9.2025.
Visits: 602 *