This study proposes an analytical framework for examining the risk factors associated with crane accidents in construction by integrating the human factors analysis and classification system (HFACS), text mining, and network visualization. An extended model, C-HFACS, was developed to systematically encode human errors and organizational influences, which were then used as structured input features for a genetic algorithm–optimized support vector machine to classify accident types. On this basis, the TF-IDF algorithm was applied to extract salient textual terms and construct feature correlation networks that clarify interrelationships among diverse accident causes. Using 130 officially published accident reports as a demonstration data set, the framework revealed that accidents were particularly frequent during disassembly and installation phases, often linked to inadequate handling of key components. The findings show that combining structured human factor coding with unstructured text mining provides a scalable approach to understanding accident causation. This study contributes to the body of knowledge by introducing a generalizable methodology that integrates human factor analysis with predictive modeling and text mining, offering practical insights for enhancing crane safety management.
Zhou et al. (2026) studied this question.