Zheng et al. (2019) Dataset
收藏arXiv2025-09-30 收录
官方服务：
资源简介：
该数据集是一个大规模的数据集，它由中文金融文档构建而成，专注于文档级别的事件提取。该数据集包含五种事件类型，涉及35种不同的论元角色。数据集按照标准划分，训练集、验证集和测试集分别包含25,632、3,204和3,204篇文档。其规模达到32,040篇文档，任务定位于文档级别的事件提取。
This is a large-scale dataset constructed from Chinese financial documents, focusing on document-level event extraction. The dataset encompasses five event types and covers 35 distinct argument roles. It is standardly split into training, validation and test sets, which contain 25,632, 3,204 and 3,204 documents respectively. The total number of documents in the dataset reaches 32,040, and the task is targeted at document-level event extraction.
