PeKo

PeKo数据集是由石溪大学创建的一个大规模标注的预条件关系数据集，包含28,948对新闻事件的预条件关系标注。该数据集通过众包方式构建，旨在帮助模型理解和推理文本中事件的必要预条件。PeKo数据集主要应用于文本分类和生成任务，特别是在需要理解事件逻辑关系和提高语言生成系统全局连贯性的场景中。

The PeKo Dataset is a large-scale annotated precondition relation dataset developed by Stony Brook University. It contains 28,948 annotated precondition relation pairs of news events. Constructed through crowdsourcing, this dataset is designed to help models understand and reason about the necessary preconditions of events within text. The PeKo Dataset is primarily applied to text classification and generation tasks, particularly in scenarios requiring comprehension of logical relationships between events and enhancement of the global coherence of language generation systems.