Abstract
Human trafficking represents the second most profitable criminal activity in the world. Here, based on the snowball sampling method, we obtained a novel dataset of a human trafficking network on the southern border of Mexico. This dataset was used to construct an unweighted and undirected graph that represents the interactions of the trafficking network. Our analysis reveals a moderate level of centralization at 44.32% and a medium density of 0.401, indicative of a structural balance that facilitates the coordination of criminal activities without a single actor’s dominance. Addressing the challenge posed by the network’s minimal cohesiveness, which hampers the sharing of resources among members, we assess four dismantling strategies: random removal, targeting hubs and brokers, a human capital-focused approach, and Generalized Network Dismantling (GND). Our findings underscore the efficacy of targeting moderately connected actors, a strategy that disrupts the network’s resilience and operational capacity by severing important but inconspicuous connections, thereby destabilizing the network’s efficiency subtly and avoiding immediate alert to the dismantling activities. This work is a significant contribution to the field of criminal network modeling and analysis.