AbstractMany human errors occur in hydraulic engineering construction, and these errors may lead to huge financial losses. A systematic and comprehensive accident analysis is required to reduce the probability of human error. Human error analysis is a lengthy and challenging process because the tendency is for accident data to be presented in text format. In addition, construction human error management requires an intelligent and efficient analysis system to ensure the timeliness of accident prevention and control. Thus, this study proposes a human error intelligent analysis system on the basis of text mining to automatically extract text knowledge and reveal the accident evolution process. Using hydraulic engineering construction text, a topic feature extraction model is built to extract words and improve the human factors analysis and classification system (HFACS) model. Then, a human error causation network that integrates text topic features, the improved HFACS model, and Bayesian theory is developed to intelligently identify human factors and quantify the human error evolution process. The analysis system proposed in this paper provides an effective way to mine and apply the experience-based knowledge available in hydraulic engineering construction text for the intelligent analysis and prediction of human error, thus improving the efficiency of human error management.