Research on the Causations of Major Construction Accidents Using Text Mining and Network Theory

The frequent occurrence of major accidents highlights the significance of studying accident reports as a pathway to accident prevention. While accidents are typically the result of multiple causes, there has been limited exploration from a network perspective regarding the interrelationships among causal factors. This study addressed this gap by constructing a sophisticated network model using 255 cases of major construction accidents in China. Through the application of text mining techniques in Python, the research identified 21 key factors across four dimensions: construction workers, construction environment, construction materials and equipment, and construction management. Subsequently, a complex network was established based on the causal links among these key factors. By analyzing topological parameters, including the degree, betweenness centrality, and average path length, this research identified illegal operations, insufficient safety education, and insufficient safety inspection as critical causal factors. This investigation offers valuable insights for accident causation analysis and advances preventive measures for enhancing construction safety.