- : Ms Word Format
- : Pages
- : ₦3000
- : 1-5 Chapters
- Click to DOWNLOAD Materials
“Advancements in Bayesian Model for Covert Member Detection in Criminal Networks via Telecommunication Metadata”
Abstract: Crime has emerged as a significant global challenge in recent times, demanding more than just a military war-fare approach to be effectively addressed. Criminal intelligence plays a crucial role in gathering data on criminal activities and participants to devise effective strategies and interventions. Social Network Analysis (SNA) has provided valuable tools for analyzing Organized Criminal Groups (OCGs) by identifying important nodes with conspicuous relationships. However, this method has limitations, as it tends to overlook silent key players while focusing on well-connected nodes.
To address these limitations, we propose a Scatter-graph of vulnerability and strategic positions that mitigates the unrelatedness of SNA metrics for the detection of key players in Criminal Social Networks (CSN). This approach identifies actors with both high vulnerability and high strategic position values, similar to the concept of Influence Maximization (IM) in which nodes with high influence are sought. However, the challenge of identifying silent key players or legitimate actors in adversary networks remains unresolved.
To address this challenge, we introduce the Missing Node concept, which focuses on nodes not initially known to be part of a criminal group but have a high affinity for well-connected nodes. Our Node Discovery scheme unravels the latent structure behind key players within CSNs. Despite its strengths, this approach fails to capture legitimate actors within a criminal group.
To overcome these limitations and improve the prediction of key players, we developed an Enhanced Bayesian Network Model (EnBNM). This model leverages the Enhanced Bayesian Model and the Recursive Bayesian Filter (RBF) algorithm to lower error rates and improve prediction accuracy. EnBNM re-ranks participants’ attributes based on the conditional probability of the Bayesian model.
We validated the EnBNM algorithm using ground truth data and adopted the SNA-Q model to classify Criminal Profile Status (CPS). The algorithm was tested using datasets from the November 17 Greece revolutionary group (N’17) and the September 11 Al-Qaeda terrorist group (9/11). The results demonstrated the effectiveness of EnBNM in detecting alleged and convicted leaders, marginal actors, and fugitives within criminal networks.
The findings also shed light on the self-organized nature of terrorist organizations, with decentralized key players to minimize the impact of security perturbations. The simulation results showed a 40% error in the court’s judgment of the N’17 group, as EnBNM detected additional key players. This underscores the importance of intelligence support in effectively disrupting OCGs and terrorist activities.
Furthermore, EnBNM displayed over 80% accuracy in detecting legitimate actors and a 59.09% accuracy score in identifying conspirators within the 9/11 terrorist group. Overall, this research highlights the critical role of intelligence in combating crime and terrorism and the significance of EnBNM in improving key player detection in criminal social networks.
Advancements in Bayesian Model for Covert Member Detection in Criminal Networks via Telecommunication Metadata”