New LAMUS corpus advances legal argument mining with LLM-assisted annotation

Researchers have introduced LAMUS, a new large-scale corpus designed for legal argument mining from U.S. caselaw. This dataset, constructed using a pipeline that combines LLM-based annotation with human refinement, focuses on sentence-level classification of judicial reasoning components. Experiments show that chain-of-thought prompting significantly enhances LLM performance on this task, while domain-specific models offer more stable zero-shot results. The LAMUS corpus aims to advance legal NLP research by providing a scalable resource and valuable empirical insights. AI

IMPACT Provides a new resource and insights for developing more sophisticated AI tools for legal analysis.

RANK_REASON The cluster describes a new academic paper introducing a dataset and methodology for legal argument mining. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• arXiv
• GitHub
• LAMUS
• Lavanya Pobbathi
• LegalBERT
• LLMs
• Supreme Court of the United States
• Texas
• U.S. caselaw

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →