This paper presents a novel approach for automated semantic analysis of legal documents using large-scale language models (LLMs). This approach aims to transform legal text into a formal representation of rebuttable deontic logic (DDL). We propose a structured pipeline that decomposes complex normative language into atomic fragments, extracts deontic rules, and assesses syntactic and semantic consistency. We evaluate the methodology against various LLM configurations, including prompt engineering strategies, fine-tuned models, and multi-stage pipelines, focusing on legal norms from the Australian Communications Consumer Protection Act. Experimental results demonstrate promising agreement between machine-generated and expert-written formalizations, demonstrating that effectively prompted LLMs can significantly contribute to scalable legal informatics.
