metadata
license: cc-by-nc-4.0
task_categories:
- text-classification
- text-generation
language:
- zh
tags:
- legal
size_categories:
- 10K<n<100K
configs:
- config_name: default
data_files:
- split: train
path:
- AppealCase/AppealCase.json
AppealCase: A Dataset and Benchmark for Civil Case Appeal Scenarios
⚖️ Introduction
The AppealCase dataset is the first large-scale resource specifically designed to support LegalAI research in appellate judgment scenarios. While prior work in LegalAI has focused heavily on one-shot trials, the appellate procedure—critical to ensuring fairness and correcting judicial errors—remains largely underexplored.
To bridge this gap, we construct 10,000 pairs of matched first-instance and second-instance civil cases from China Judgments Online, covering 91 causes of action. We also design a structured annotation schema and define new tasks tailored to the appellate context.
📰 News
- [2025/05/23] 🎉 We have released our arXiv paper and the accompanying dataset!
📂 Dataset
- Cases: 10,000 matched pairs of first-instance and second-instance judgments
- Coverage: 91 civil causes of action
- Labeling: Annotated with judgment reversal, reasons for reversal, claims, legal provisions and new information
- Format: JSON structured format, easy to load and parse
- License: CC BY-NC 4.0
📚 Citation
@article{huang2025appealcase,
title={{AppealCase}: A dataset and benchmark for civil case appeal scenarios},
author={Huang, Yuting and Guo, Meitong and Wu, Yiquan and Li, Ang and Liu, Xiaozhong and Yin, Keting and Sun, Changlong and Wu, Fei and Kuang, Kun},
journal={arXiv preprint arXiv:2505.16514},
year={2025}
}