“Progressing the Enhanced Bayesian Model for Detecting Covert Members within Criminal Networks via Telecommunication Metadata Analysis



Crime poses an increasingly complex global challenge that cannot be effectively addressed solely through traditional military warfare approaches. Intelligence gathering is crucial in combating criminal activities, especially within Organised Criminal Groups (OCGs). Social Network Analysis (SNA) offers valuable tools for analyzing these criminal networks, focusing on identifying key players and their relationships. However, SNA has limitations, as conspicuous links can expose active participants while concealing silent key actors. Additionally, the status of key actors in OCGs is not necessarily related to SNA metrics.

To overcome these limitations, we propose a Scatter-graph of vulnerability and strategic positions to detect key players in the Criminal Social Network (CSN). This approach identifies actors with both high vulnerability and strategic position values, akin to Influence Maximization (IM) techniques. However, it still leaves silent key players or legitimate actors in the adversary network unresolved.

To address this, we introduce the Missing Node concept, which focuses on nodes initially unknown to be part of the social criminal group. It prioritizes well-connected nodes over marginal ones. Furthermore, our Node Discovery scheme uncovers the latent structure behind key players in the CSN, although legitimate actors may not be fully captured.

To enhance the prediction of key players, such as financial aiders and ammunition suppliers with evasive behaviors, we developed the Enhanced Bayesian Network Model (EnBNM). This model combines the Bayesian model and Recursive Bayesian Filter (RBF) algorithm to reduce error rates and improve predictions.

We validated the EnBNM algorithm using ground truth data and adopted the SNA-Q model for classifying Criminal Profile Status (CPS). Testing it on datasets from the N‟17 and 9/11 groups, EnBNM successfully detected alleged and convicted leaders, marginal actors, and fugitives. Moreover, it identified additional key players not detected by previous models, demonstrating the significance of intelligence support in disrupting OCGs and terrorist organizations.

The simulation results reveal that relying solely on court judgments may lead to errors, underscoring the need for intelligence assistance in combating crime and terrorism effectively. EnBNM also proved adept at detecting legitimate actors and conspirators within terrorist groups, further emphasizing its potential impact in combating criminal activities and enhancing security efforts.

Progressing the Enhanced Bayesian Model for Detecting Covert Members within Criminal Networks via Telecommunication Metadata Analysis. GET MORE, ACTUARIAL SCIENCE PROJECT TOPICS AND MATERIALS