Hazard identification is a critical step in both safety and security risk assessments. Traditional approaches predominantly rely on historical data and Exploratory Data Analysis (EDA) to define reference scenarios. However, EDA lacks standardized methods for identifying and ranking incident chains. To overcome this limitation, the BAS2E (Bayesian network Analysis of Safety and Security Events) methodology was developed to systematically derive reference scenarios from historical event data using Bayesian Networks (BN). The methodology uses statistical data from accident records and applies the Noisy-OR gate model to manage uncertainties in the specification of conditional probability tables (CPTs). Sensitivity analysis is used to quantify the influence between nodes in the BN, enabling a ranked selection of the most critical incident chains for inclusion in risk assessment. The methodology is demonstrated using a dataset of 109 security incidents occurred in the offshore Oil&Gas industry.
Matteo et al. (Wed,) studied this question.