Precedent-Enhanced Legal Judgment Prediction with LLM and Domain-Model Collaboration
Yiquan Wu1, Siying Zhou1, Yifei Liu1, Weiming Lu1∗, Xiaozhong Liu2 Yating Zhang3, Changlong Sun13, Fei Wu1∗, Kun Kuang1∗ 1Zhejiang University, Hangzhou, China 2Worcester Polytechnic Institute, USA 3Alibaba Group, Hangzhou, China {wuyiquan, zhousiying, liuyifei, luwm,
Abstract
Legal Judgment Prediction (LJP) has become an increasingly crucial task in Legal AI, i.e., predicting the judgment of the case in terms of case fact description. Precedents are the previous legal cases with similar facts, which are the basis for the judgment of the subsequent case in national legal systems. Thus, it is worthwhile to explore the utilization of precedents in the LJP. Recent advances in deep learning have enabled a variety of techniques to be used to solve the LJP task. These can be broken down into two categories: large language models (LLMs) and domain-specific models. LLMs are capable of interpreting and generating complex natural language, while domain models are efficient in learning task-specific information. In this paper, we propose the precedent-enhanced LJP framework (PLJP) – a system that leverages the strength of both LLM and domain models in the context of precedents. Specifically, the domain models are designed to provide candidate labels and find the proper precedents efficiently, and the large models will make the final prediction with an in-context precedents comprehension. Experiments on the real-world dataset demonstrate the effectiveness of our PLJ
中文速览
法院判决预测（Legal Judgment Prediction）长期面临一个难题：如何像真实法官那样参考先例来做出准确裁判。本文提出了一个名为 PLJP 的先例增强判决预测框架，巧妙地让领域专用模型（domain-specific model）和大语言模型（Large Language Model, LLM）各司其职、协同工作——领域模型负责从海量案例库中高效筛选候选标签并检索最相关的先例，LLM 则以这些先例为上下文进行理解并做出最终预测。为避免 LLM 训练数据与测试集重叠导致的数据泄露，研究者额外构建了一个涵盖 2022 年后案件的新测试集，在两个测试集上 PLJP 均达到了最优性能。这项工作不仅推动了法律 AI 的发展，还为 LLM 与领域模型协作这一思路提供了可推广到其他垂直领域的范式参考。
原文 arXiv:2310.09241;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2310.09241v1