Identification of risk factors for coal mine accidents based on text mining and social network analysis

In response to the coal mining industry's high-risk nature and limitations of traditional accident analysis, this study constructs a multi-factor coupling analysis framework using 481 accident reports. Parsing unstructured text reveals 'core–periphery' structural characteristics in accident causation systems. Key contributions of the study are as follows: methodologically, it employs text mining to automate factor extraction and integrates social network analysis (SNA) to quantify node centrality and transmission intensity; theoretically, 18 core causations (e.g., unauthorized risk-taking) are network hubs, while 50 peripheral factors (e.g., latent equipment defects) amplify core risks through linkages, validating 'minor signals triggering major accidents' dynamics; and practically, targeted critical node intervention strategies are proposed, aiding a shift from single-factor control to networked management and offering global high-risk industry insights.