ASER

该数据集名为ASER，是一个基于事件的大型知识图谱，它通过语言图上的选择偏好来捕捉常识知识。该数据集不仅包含了知识图谱的访问接口和工具，其规模宏大，拥有6.48亿条边连接4.38亿个事件实体。在概念化处理后，数据集包含了1500万个概念化事件实体和2.24亿条边。这一数据集的任务旨在进行常识知识的获取和推理。

ASER is a large-scale event-based knowledge graph dataset that captures commonsense knowledge via selectional preferences over linguistic graphs. It not only provides access interfaces and supporting tools for the knowledge graph, but also boasts an enormous scale, with 648 million edges connecting 438 million event entities. After conceptualization, the dataset contains 15 million conceptualized event entities and 224 million edges. The tasks supported by this dataset aim at commonsense knowledge acquisition and reasoning.