CJRC

CJRC是中国首个司法阅读理解数据集，由联合实验室HIT和iFLYTEK（HFL）创建，包含约10,000份判决文件和近50,000个问题及答案。数据集内容主要来源于中国的判决文书，由法律专家进行标注，涵盖民事和刑事案件，涉及多种罪名和诉讼原因。创建过程中，专家们针对事实描述部分标注了四到五个问题-答案对，包括不可回答和是/否类型的问题。该数据集主要用于法律领域的元素提取，帮助研究人员通过阅读理解技术快速提取信息，提高法官的工作效率和决策质量。CJRC is the first judicial reading comprehension dataset in China, developed by the joint laboratory HIT and iFLYTEK (HFL). It contains approximately 10,000 judgment documents and nearly 50,000 question-answer pairs. The dataset is primarily sourced from Chinese judicial documents, annotated by legal experts, and covers civil and criminal cases involving a wide range of criminal charges and causes of action. During its development, experts annotated 4 to 5 question-answer pairs for each factual description section, including unanswerable and yes/no type questions. This dataset is mainly utilized for legal domain element extraction, assisting researchers in rapidly extracting information via reading comprehension technologies, and improving judges' work efficiency and decision-making quality.