LawArgueAgent: A Framework to Enhance Legal Judgment Prediction via Lawyer-Adversarial Self-Play and Case Generation

ACL ARR 2026 January Submission1696 Authors

Keywords: agent evolution, knowledge augment, RAG

Abstract: Given the fact of a legal case, Legal Judgment Prediction (LJP) aims to make judicial outcomes, including relevant legal charge, terms, and fines. However, LJP faces two key challenges: (1)Long-Tail Distribution: Existing datasets, derived from authentic cases, suffer from high human annotation costs and imbalanced distributions, leading to model performance degradation. (2)Lack of Lawyer Augmentation: Current systems focus on enhancing judges' decision-making but neglect the critical role of lawyers in refining argument, thereby limiting overall judicial accuracy. To address these issues, we propose LawArgueAgent, an adversarial self-play lawyer augmented legal judgment framework, which integrates a case generation module to tackle long-tailed data distributions and an adversarial self-play mechanism to enhance lawyers' argumentation skills. Our experiments on a Chinese legal dataset show that our framework enables a weak model, Qwen1.5-7B-Chat, to surpass powerful models like GPT-4 in legal judgment prediction. This demonstrates the effectiveness of our approach in improving LJP performance by simulating a courtroom adversarial process.

Paper Type: Long

Research Area: NLP Applications

Research Area Keywords: NLP Applications,Resources and Evaluation

Contribution Types: NLP engineering experiment

Languages Studied: Chinese

Submission Number: 1696