Leveraging Natural Language Processing for hazard pattern discovery: A decadal study of severe chemical accidents in China (2011–2023)

Accident reports constitute vital carriers of specialized knowledge, serving as the primary medium for industrial safety information. Given the time‐consuming nature of manual report analysis, this investigation examines severe chemical accidents in China (January 2011 to March 2023) using text mining techniques. Through collection and analysis of 115 official accident reports, we extracted critical information encompassing incident chronologies, causation, regulatory sanctions, and prevention strategies. The analysis revealed three key findings: (1) spatiotemporal distribution patterns of accidents, (2) underlying risk determinants specific to China's process industries, and (3) systemic vulnerabilities in organizational safety management systems. This study further proposes data‐driven countermeasures for process safety enhancement. The evidence‐based conclusions offer scientific rationale for optimizing chemical safety governance systems and implementing precision interventions to reduce accident probabilities.