Dynamic Heterogeneous Academic Graphs

该数据集是一组大规模、动态异构的学术图表集合，涵盖了从人工智能（AI）和核不扩散（NN）领域的科学出版物中提取的上下文和内容信息。这些数据集由专家针对特定主题精心筛选关键词，并通过使用特定工具从多种来源中提取出版物元数据。在规模上，该数据集包含了来自顶级会议的大型AI数据集以及拥有超过50万个节点的核数据集。其任务旨在解决多步骤的图表预测问题。

This dataset is a large-scale, dynamic heterogeneous collection of academic graphs, which encompasses contextual and content information extracted from scientific publications across the fields of artificial intelligence (AI) and nuclear non-proliferation (NN). Curated by experts who meticulously select keywords for specific research topics, the dataset extracts publication metadata from diverse sources using specialized tools. In terms of scale, it comprises large-scale AI datasets sourced from top-tier conferences, as well as a nuclear dataset with more than 500,000 nodes. The core task of this dataset focuses on solving multi-step graph prediction problems.