A Vision for Automated Building Code Compliance Checking by Unifying Hybrid Knowledge Graphs and Large Language Models

In the Architecture, Engineering, and Construction (AEC) industry, ensuring compliance with building regulations is a complex but necessary task, traditionally done manually with a high risk of error. Automated Compliance Checking (ACC) aims to improve efficiency and accuracy by using digital tools to compare building designs against codes, but current ACC solutions still face limitations in automating the interpretation of regulations and seamlessly integrating them with building data. This paper introduces a novel framework for automating building code compliance checking within the AEC industry. The proposed approach unifies Hybrid Knowledge Graphs (HKGs) and Large Language Models (LLMs) to overcome limitations in existing methods. Building information and regulations are integrated within a hybrid knowledge graph structure, while LLMs are leveraged to access, infer, and encode this complex data for automated compliance checks. The paper outlines a detailed methodology encompassing Knowledge Graph (KG) construction, LLM integration, and evaluation strategies. A case study demonstrates feasibility, with results highlighting the potential of this framework as well as areas for further refinement. This research offers a promising foundation for enhanced efficiency, accuracy, and scalability in automated building code compliance, with the potential to transform practices within the AEC industry.