How Can ChatGPT Help in Automated Building Code Compliance Checking?
Jiansong Zhang
Abstract
How Can ChatGPT Help in Automated Building Code Compliance Checking? Jiansong Zhang Pages 63-70 (2023 Proceedings of the 40th ISARC, Chennai, India, ISBN 978-0-6458322-0-4, ISSN 2413-5844) Abstract: One main challenge in the full automation of building code compliance checking is in the extraction and transformation of building code requirements into computable representations. Semantic rule-based approach has been taken mainly due to its expected better performance than machine learning-based approach on this particular task. With the recent advancement in deep learning AI, particularly the launch of ChatGPT by OpenAI, there is a potential for this landscape to be shifted given the highly regarded capabilities of ChatGPT in processing (i.e., understanding and generating) natural language texts and computer codes. In this paper, the author preliminarily explored the use of ChatGPT in converting (i.e., extracting and transforming) building code requirements into computer codes, and compared it with the results from cutting-edge semantic rule-based approach. It was found that comparing to the semantic rule-based approach, the conversion results from ChatGPT still has limitations, but there is a great potential for it to help speed up the implementation and scale-up of automated building code compliance checking systems. Keywords: Automated building code compliance checking, AI, ChatGPT, SNACC, Natural language processing DOI: https://doi.org/10.22260/ISARC2023/0011 Download fulltext Download BibTex Download Endnote (RIS) TeX Import to Mendeley Presentation Video: https://youtu.be/cLXvIm9s2R0