Treffer: Development of a Bayesian Network-based Safety Performance Quantification Model on building construction projects in Korea.

As the integration of digital technologies in construction sites advances, interest in utilizing unstructured data is surging. In particular, the importance of safety inspections and proactive measures utilizing the vast amount of field-generated documentation is growing. The Korean construction industry has recently strengthened its systems and standards for preventing safety accidents. However, there are limitations in evaluating and improving the performance using only structured data based on existing institutional standards. This study proposes a Safety Performance Quantification Model (SPQM) to assess safety performance by utilizing the rapidly increasing unstructured data. The SPQM employs a Bayesian Network to evaluate safety performance through natural language processing techniques and analysis of association rules and social networks. The SPQM also leverages safety inspection documents produced by site supervisors to train Bayesian networks with Python 3.8 and verify network performance through the Brier Score (BS). The BS of the trained model is below 0.25, and the prediction rate is approximately 80%. The Bayesian Network-based SPQM can be a decision-making tool to quantify performance using unstructured data. In the future, the SPQM is expected to improve the timeliness of response by monitoring safety performance in conjunction with institutionally required data analysis. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Asian Architecture & Building Engineering is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)