From natural language to interoperable BIM: Extensible framework for compliance-aware generative design

In design tasks, ensuring that architectural layouts comply with regulations is essential for design quality, usability, and legitimation. However, current workflows usually treat design and compliance checking as separate stages, resulting in lengthy and potentially repetitive checking and revision cycles. To address that, this paper presents an extensible framework to integrate design knowledge from both provisions and natural-language design intents and generate compliance-aware layouts. The framework:(1) represents design knowledge as ontology-grounded constraints, (2) applies the constraints in topology generation, and (3) subsequently compiles generated layouts into interoperable Building Information Modeling (BIM) instance models. The framework is demonstrated through a case study. The generated layouts are validated across different BIM environments. Compliance rate, intent alignment, and time efficiency are evaluated against manual workflows. Extensibility tests further show that the proposed framework can accommodate additional provisions, more detailed intents, and new semantic concepts. The results indicate that the proposed framework provides a solution for efficient compliance-aware generative design.