Julian2002/pdp4k
收藏资源简介:
PDP-4K 是一个中文检察机关公诉决定预测数据集,包含 4,730 条结构化检察文书样本。数据集的任务是给定犯罪嫌疑人信息、案件程序和案件事实,预测适用法条和公诉决定。适用法条使用 `cl:`、`cpl:`、`cpr:` 等前缀标识法律来源,公诉决定为四分类标签,包括起诉、相对不起诉、法定不起诉、存疑不起诉。数据集的结构包括多个字段,如id、meta、person_info、procedure、fact、relevant_articles、decision等。数据集还提供了详细的标签说明和注意事项,用于法律 NLP 研究与模型评测,不构成法律意见。
PDP-4K is a Chinese prosecution decision prediction dataset containing 4,730 structured prosecution document samples. The task of the dataset is to predict applicable legal articles and prosecution decisions given suspect information, case procedures, and case facts. The applicable legal articles are identified using prefixes such as `cl:`, `cpl:`, and `cpr:`, and the prosecution decisions are four-class labels including prosecution, relative non-prosecution, statutory non-prosecution, and doubtful non-prosecution. The dataset structure includes multiple fields such as id, meta, person_info, procedure, fact, relevant_articles, decision, etc. The dataset also provides detailed label descriptions and precautions for legal NLP research and model evaluation, and does not constitute legal advice.




