Subway construction accident reports often take a lot of time and personnel to analyze and contain a large amount of data that require professional identification, which increases the difficulty of the analysis. This study aims to use Generative Pre-trained Transformer (GPT) models for the automated analysis of subway construction accident investigation reports, with the goal of improving the efficiency of accident identification and analysis. By analyzing a dataset of 50 subway reports, this study developed the Accident Investigation Report (AIR) Agent, which utilizes GPTs to automatically identify accident types and extract key details from the reports. The chatbot is composed of three core modules: a conversation module, an instruction module, and a knowledge module. Ablation studies were performed to validate the AIR Agent’s efficiency, and the validation results show that the AIR Agent achieves an accuracy of 80.32% when analyzing new reports with a brief conclusion, demonstrating the AIR Agent’s ability to automatically format and structure reports in a consistent and correct manner. These findings suggest that the AIR Agent can significantly reduce the manual effort involved in accident investigation report analysis and enhance the overall efficiency of analyzing subway construction accident investigation reports, thereby improving the effectiveness of accident investigation and management.

The construction industry continues to face challenges such as increased costs, time overruns, and low quality. Off-site construction (OSC) methods are increasingly being adopted as alternatives to traditional construction practices to address these issues, with off-site manufacturing (OSM) representing a key difference in construction methods. However, existing studies have largely neglected the systematic evaluation of OSM risks on quality, cost, and delivery (QCD) outcomes, leaving a significant gap in understanding the complex interdependencies among risk factors. To improve risk management in OSC projects, it is crucial to evaluate the impact of OSM risks on QCD outcomes. This study applies the Bayesian Belief Network (BBN) method to develop an evaluation model that measures the impact of OSM risks on QCD outcomes in OSC projects. The results identify 12 significant risk factors affecting QCD outcomes in OSC projects. Five key risk groups were identified as critical for managing OSM risks. This approach provides a systematic framework for managing OSM risks and optimizing OSC practices in China.

Unstructured aviation accident narratives are difficult to convert into causal, decision-oriented models. This research presents and validates a generative pre-trained transformer (GPT)-powered automated analysis framework for aviation safety risk. GPT models systematically analysed 23,620 unstructured accident reports from the Aviation Safety Network (ASN) using a templated prompt to extract cause events, identifying 12 effect events, 33 cause events, and their probabilistic dependencies. A Bayesian belief network (BBN) was developed through a simulated annealing structure search guided by the Bayesian Dirichlet equivalent uniform (BDeu) score, and its parameters were estimated through maximum likelihood estimation. The model was validated through five-fold cross-validation, achieving an overall accuracy of 86.58%. Tornado-based sensitivity analysis and composite risk ranking identified aircraft design defect, crew disregard of regulations, dust haze, hydraulic failure, and thunderstorms as the five most critical cause events, forming the basis for targeted risk-control recommendations. This research contributes a GPT-driven extraction and mapping workflow that turns narratives into cause events, and is deployed as a BBN for aviation risk inference on 23,620 ASN accidents with recommendations. This research shows that the GPT-powered extraction process supports systematic risk analysis; on an expert-adjudicated benchmark, it achieved higher recall, micro-F1, and macro-F1 than keyword and TF–IDF baselines, while the BBN translated the extracted events into risk-control priorities.