Isah, M A and Kim, B S (2025) Question-answering system powered by knowledge graph and generative pretrained transformer to support risk identification in tunnel projects. Journal of Construction Engineering and Management, 151(1): 04024193, ISSN 0733-9364
Abstract
Risk identification is fundamental to effective risk management in any construction project. It is especially true for tunnel projects where technicality and complexity increase the risks, uncertainties, and challenges of realistic risk identification. However, two important issues need to be addressed in the current risk identification practice. First is the organization of previous project risk information (data) in unstructured and semistructured formats which hinders the use of emerging AI-based techniques and digitization of the risk management process. Second, overreliance on manual efforts and human experience, which is labor-intensive, error-prone, and time-consuming. To address these issues, this study proposes an intelligent question-answering system based on a knowledge graph and generative pretrained transformers (GPT) model. This study developed a tunnel risk knowledge graph (TRisKG) and then integrated the TRisKG with a GPT model to develop a question-answering system for tunnel project risks (QASTRisk). The proposed QASTRisk yielded a precision value of 97%, recall of 94%, and an F-1 score of 95%, indicating excellent performance compared to the existing studies. The proposed approach is appropriate for quick, effective, and intuitive risk identification at the early stage of tunnel projects as it can facilitate timely and reliable decision-making to prevent safety accidents and reduce project delays and cost overruns. Also, this work contributes to the body of knowledge by providing a framework for developing intelligent systems using artificial intelligence (AI) techniques to assist project managers in automatic and efficient risk identification at the preconstruction phase of the project. Thus, this enhances decision-making, improves project performance and the efficiency of project risk management practices, and increases productivity in the construction industry.
| Item Type: | Article |
|---|---|
| Uncontrolled Keywords: | generative pretrained transformer (gpt) model; knowledge graph; question-answering system; risk identification; risk management; tunnel projects |
| Index terms: | project manager, efficiency, project risk management, intelligent system, cost overrun, productivity, project performance, risk management, artificial intelligence, construction industry, tunnel, risk identification, construction project, decision-making, project delay, complexity, body of knowledge |
| Subjects: | artificial intelligence, financial and cost management, risk assessment, profession, project controls, decision analysis, industry analysis, infrastructure and transport systems, systems engineering, management, performance management, project management theory and practice, knowledge management, automation and robotics, production management, financial risk |
| Topics: | Project Management, Engineering Principles, Risk Management, Quality Management, Research Practice, Information Management, Cost Management, Business Strategy, Roles and Professions, Digital Applications, Time Control |
| Descriptive scope: | 3 PCT |
N.B. Descriptive scope is a count of how many of the five facets of empirical research are indicated by the words used in title, abstract and keywords. It is not intended as a judgement on the research; merely a count of the kind of word we would expect to indicate Phenomenon, Concepts, Theoretical framing, Empirical techniques, Analytical techniques. If all five are present, then a code of “5 PCTEA” will indicate this. If you feel the coding for this record is questionable, we welcome discussion around the terms we matched or the way we categorized them. The facet you would expect may not be coded, or a facet may be coded inappropriately. This can also bear on a larger question, of which facets should be treated as defining in construction management research. Please get in touch, and we will look at it. More details here