Enhancing maritime transportation security: A data-driven Bayesian network analysis of terrorist attack risks

Maritime terrorist accidents have a significant low-frequency-high-consequence feature and, thus, require new research to address the associated inherent uncertainty and the scarce literature in the field. This article aims to develop a novel method for maritime security risk analysis. It employs real accident data from maritime terrorist attacks over the past two decades to train a data-driven Bayesian network (DDBN) model. The findings help pinpoint key contributing factors, scrutinize their interdependencies, ascertain the probability of different terrorist scenarios, and describe their impact on different manifestations of maritime terrorism. The established DDBN model undergoes a thorough verification and validation process employing various techniques, such as sensitivity, metrics, and comparative analyses. Additionally, it is tested against recent real-world cases to demonstrate its effectiveness in both retrospective and prospective risk propagation, encompassing both diagnostic and predictive capabilities. These findings provide valuable insights for the various stakeholders, including companies and government bodies, fostering comprehension of maritime terrorism and potentially fortifying preventive measures and emergency management.