Legal Fact Prediction: The Missing Piece in Legal Judgment Prediction

2024-09-11

AI Summary

To address the practical challenge in legal judgment prediction (LJP) caused by the absence of adjudicated legal facts during early litigation stages, this paper introduces a novel task—Legal Fact Prediction (LFP): predicting judicially affirmed legal facts from raw evidence submitted by litigants. We formally define LFP for the first time, construct LFPBench—the first benchmark dataset for this task—and propose a multi-stage end-to-end framework integrating evidence extraction, fact structuring, and case-law alignment, enhanced by large language model fine-tuning and retrieval-augmented generation. Experiments demonstrate that our method achieves significant improvements in prediction accuracy on LFPBench. Moreover, integrating LFP into downstream LJP boosts F1-score by 12.7%. To foster transparency and real-world deployment of legal AI, we publicly release both the code and the LFPBench dataset.

Technology Category

• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Machine Learning: Large Multimodal Models (LMMs)
• Reasoning under Uncertainty: Other Foundations of Reasoning under Uncertainty

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Legal judgment prediction (LJP), which enables litigants and their lawyers to forecast judgment outcomes and refine litigation strategies, has emerged as a crucial legal NLP task. Existing studies typically utilize legal facts, i.e., facts that have been established by evidence and determined by the judge, to predict the judgment. However, legal facts are often difficult to obtain in the early stages of litigation, significantly limiting the practical applicability of fact-based LJP. To address this limitation, we propose a novel legal NLP task: extit{legal fact prediction} (LFP), which takes the evidence submitted by litigants for trial as input to predict legal facts, thereby empowering fact-based LJP technologies to perform prediction in the absence of ground-truth legal facts. We also propose the first benchmark dataset, LFPBench, for evaluating the LFP task. Our extensive experiments on LFPBench demonstrate the effectiveness of LFP-empowered LJP and highlight promising research directions for LFP. Our code and data are available at https://github.com/HPRCEST/LFPBench.

Problem

• Predict legal facts from evidence in early litigation stages.
• Enhance legal judgment prediction without ground-truth legal facts.
• Introduce LFPBench dataset for evaluating legal fact prediction.

Innovation

• Introduces legal fact prediction (LFP) task
• Uses evidence to predict legal facts
• Creates LFPBench benchmark dataset