overthelex/ukrainian-court-decisions
收藏资源简介:
这是一个乌克兰法院判决数据集,用于案件判决预测任务。数据来源于乌克兰国家法院判决注册表(ЄДРСР),包含乌克兰语的民事和商业法院判决,时间跨度为2003年至2025年。数据集共150,000个样本,平衡分为三个类别:approved(完全支持)、dismissed(驳回)和partial(部分支持)。输入为法院判决的事实部分(ВСТАНОВИВ),任务是根据事实预测判决结果。数据集经过预处理,确保事实部分长度在200到10,000字符之间,整体文本长度在500到200,000字符之间。数据分为训练集(120,000样本)、验证集(15,000样本)和测试集(15,000样本),支持多种配置(如default、full_scale、hybrid_war、pre_war等),适用于法律自然语言处理研究,特别是判决预测和时间概念漂移分析。
A dataset of Ukrainian court decisions for case outcome prediction, extracted from the State Court Decisions Registry (ЄДРСР). It contains Ukrainian-language civil and commercial court decisions from 2003 to 2025. The dataset includes 150,000 samples, balanced across three classes: approved (claim fully satisfied), dismissed (claim dismissed), and partial (claim partially satisfied). The input is the facts section (ВСТАНОВИВ) of court decisions, and the task is to predict the judgment outcome based on the facts. Data is preprocessed with a minimum of 200 characters and a maximum of 10,000 characters for the facts section, and full text length between 500 and 200,000 characters. It is split into train (120,000 samples), validation (15,000 samples), and test (15,000 samples) sets, with multiple configurations (e.g., default, full_scale, hybrid_war, pre_war) for legal NLP research, particularly judgment prediction and temporal concept drift analysis.




