Large Language Model-based data-driven framework for digital transformation in construction industry

Abstract

With the wide and fragmented use of digital technology in construction, a systematic digital transformation (DT) of the industry is needed. The industry’s synergy development context, marked by diverse data resources and significant investment, complicates collaboration and burdens the DT process. Notably, the transformation knowledge of DT is often “buried” within the vast data produced by daily management processes, making it challenging to discern the rules of DT without labor-intensive and time-consuming manual methods. Hence, a well-established data-driven framework for enhancing the DT process to promote whole-lifecycle industry transformation is essential. The large language model (LLM) supercharges the data-driven framework, enabling automated reasoning and precise insights to be derived from extensive data sets, thus fostering a smarter DT framework to manage the DT process. Therefore, this study uses a question-answering system based on an LLM and a localized knowledge base to guide decision-makers in developing engagement strategies that improve DT performance and foster collaboration. This study presents a practical application of LLMs in the DT of construction enterprises, anticipates future applications, and explores their potential use throughout a construction project’s transformation lifecycle.