Modeling threats of mass incidents using scenario-based Bayesian network reasoning

Mass incidents represent a global problem, putting potential threats to public safety. Due to the complexity and uncertainties of mass incidents, law enforcement agencies lack analytical models and structured processes for anticipating potential threats. To address such challenge, this paper presents a threat analysis framework combining the scenario analysis method and Bayesian network (BN) reasoning. Based on a case library of mass incidents in China, a BN capturing the interaction of twelve key factors in mass incidents is developed, where the network structure is determined by data and expert knowledge. The model is compared with two base-line BN models (use only expert knowledge or data) and a logistic regression model, proving to be the most robust. Using sensitivity analysis, we further identify a more critical subset of those threat-predicting factors. Finally, we present a case study to demonstrate how to apply the proposed framework to assessing the threat of ongoing mass incidents.