De Jure: Iterative LLM Self-Refinement for Structured Extraction of Regulatory Rules
By: Keerat Guliani, Deepkamal Gill, David Landsman, Nima Eshraghi, Krishna Kumar, Lovedeep Gondara
Published: 2026-04-03
View on arXiv →#cs.AI
Abstract
This research presents "De Jure," a system for iterative large language model (LLM) self-refinement aimed at structured extraction of regulatory rules. It addresses the critical need for accurate and automated interpretation of complex legal and compliance documents, with significant potential for applications in legal tech and regulatory compliance.