Reproduce “Large Language Model-Driven Code Compliance Checking in Building Information Modeling” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Soumya Madireddy, Lu Gao, Zia Din, Kinam Kim, Ahmed Senouci, Zhe Han, Yunpeng Zhang

An LLM-driven framework was developed to automatically interpret building codes and generate Python scripts for real-time, semi-automated compliance checking within Autodesk Revit. This system successfully identified compliant and non-compliant elements across 12 different building rules in case studies and produced actionable reports, with Grok achieving an average script generation success rate of 76.7%.

Key insights

  • Large Language Models can effectively interpret complex natural language building codes and translate them into executable Python scripts for direct compliance checking within BIM environments.
  • Iterative and highly detailed prompt engineering is crucial for enabling LLMs to generate accurate and functional code that adheres to the specific requirements and limitations of specialized software APIs and scripting environments like Revit's IronPython.
  • The performance of various LLMs in code generation for AEC compliance checking varies significantly, with some models (e.g., Grok, Claude Sonnet 3.5) demonstrating higher success rates and requiring fewer correction attempts than others.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.