It has been revealed in numerous investigation reports that human and organizational factors (HOFs) are the fundamental causes of coal mine accidents. However, with various kinds of accident-causing factors in coal mines, the lack of systematic analysis of causality within specific HOFs could lead to defective accident precautions. Therefore, this study centered on the data-driven concept and selected 883 coal mine accident reports from 2011 to 2020 as the original data to discover the influencing paths of specific HOFs. First, 55 manifestations with the characteristics of the coal mine accidents were extracted by text segmentation. Second, according to their own attributes, all manifestations were mapped into the Human Factors Analysis and Classification System (HFACS), forming a modified HFACS-CM framework in China's coal-mining industry with 5 categories, 19 subcategories and 42 unsafe factors. Finally, the Apriori association algorithm was applied to discover the causal association rules among external influences, organizational influences, unsafe supervision, preconditions for unsafe acts and direct unsafe acts layer by layer, exposing four clear accident-causing "trajectories" in HAFCS-CM. This study contributes to the establishment of a systematic causation model for analyzing the causes of coal mine accidents and helps form corresponding risk prevention measures directly and objectively.
Keywords: HFACS framework; association rules; coal mine accidents; data-driven; text mining.