Proposes a workflow combining improved text mining, Apriori association-rule mining, and Bayesian networks to identify and connect safety risk factors from 330 chemical accident investigation reports. It is relevant to Process Safety Management because the Bayesian network structure and sensitivity analysis can support incident investigation, recurring-cause discovery, learning from experience, and risk-based prioritization of corrective actions.
Note: This paper uses a number of older statistical analysis techniques (e.g. PageRank) that are not generative AI-based. However, the amount of source data they have is impressive, and the analysis techniques are insightful. So, we've included it for psm.ai readers.
Z. Zhou, J. Huang, Y. Lu, H. Ma, W. Li, and J. Chen, "A New Text-Mining–Bayesian Network Approach for Identifying Chemical Safety Risk Factors," Mathematics, vol. 10, no. 24, Art. no. 4815, Dec. 2022, doi: 10.3390/math10244815.
