Framework for automated building code compliance checking to improve transparency, trust, validation, and design interpretation

Building codes are complex and written in natural language, making interpretation costly and prone to error. This paper addresses the specific question of how Natural Language Processing (NLP) can automate code compliance checking across entire building codes rather than isolated sections. The proposed framework uses a two-stage method that first structures code documents into semantic hierarchies and then applies ontological reasoning to generate and verify rule sets against BIM models. Results show that the approach captures exceptions and cross-references more effectively than current rule-based or commercial systems. This is important for architects and engineers who must ensure accuracy, transparency, and trust in compliance decisions while reducing manual effort. The framework also establishes a foundation for scalable, fully automated compliance systems that can guide future research on integrating artificial intelligence with building regulations. • Proposed a framework that improves transparency in automated code compliance by providing clear traceability of rules and decisions. • Enhanced trust through human-in-the-loop mechanisms that allow designers to verify exceptions, confirm rule interpretations, and oversee compliance outcomes. • Advanced validation by developing a semantic pipeline that transforms building codes into structured, machine-readable rule sets and enables cross-checking across multiple code versions. • Supported better design interpretation with a space-based ontological model that captures relationships between building components and spaces for inference-driven reasoning. • Demonstrated scalability and practical deployment by enabling compliance checking across different codes (e.g., IBC, IRC) using low-computational resources.