Causal and correlation analysis of high-rise building fires using text mining and ISM: Evidence from China

Research on high-rise building fire causal mechanisms remains limited. This study proposes a comprehensive fire cause analysis model for high-rise buildings by integrating text mining with the interpretive structural modelling (ISM) method. Based on text mining of a Chinese dataset containing 123 fire accident investigation reports from 2000 to 2024, 16 causes of high-rise building fire causes were objectively identified. Using UCINET analysis, 14 key causes were extracted, and the Apriori algorithm was applied to reveal 14 strong correlations among them. Subsequently, the ISM method was employed to structure these causes into six hierarchical levels. Results indicate that insufficient fire laws and policies, together with inadequate administrative management, are the fundamental root causes. At the uppermost level, direct causes include fire-related human activities, electrical malfunctions, and defects in fire protection facilities. The findings enhance the objectivity of fire cause identification and provide valuable insights for fire prevention and in high-rise buildings. • A novel fire cause analysis model for high-rise buildings was developed by integrating text mining. • Sixteen fire causes were identified, and fourteen key causes were further extracted. • Strong correlation relationships among the key causes were revealed using the Apriori algorithm. • Logical hierarchical relationships of fire causes were structured and interpreted through the ISM approach.