Enhancing cyber risk identification in the construction industry using language models
Dongchi Yao, Borja García de Soto
Abstract
Modern construction projects are vulnerable to cyber-attacks due to insufficient attention to cybersecurity. Cyber risks in construction projects are not fully recognized, and the relevant literature is limited. To address this gap, the capabilities of a language model were leveraged to analyze extensive text, tailored to identify cyber risks. The model was trained using a curated corpus related to construction cybersecurity, enhanced by Supervised Fine-Tuning and Reinforcement Learning from Human Feedback techniques. The findings demonstrate advancements in the model's ability to understand cybersecurity and generate responses to cybersecurity questions. Using this model, a prioritized checklist of cyber risks across project phases was developed, establishing a new industry benchmark . This checklist can be utilized by various groups, including project managers and risk analysts. The model allows for updates with new data, ensuring the checklist remains current. The upgraded model holds significant promise for industry-wide applications, serving as an intelligent cybersecurity consultant.