De Jure: Iterative LLM Self-Refinement for Structured Extraction of Regulatory Rules | ScienceToStartup | ScienceToStartup