andreas-helgesson/gdelt-big

GDELT-big是一个遵循FinDKG和RE-GCN格式的时态知识图谱（TKG）数据集。它由GDELT 1.0全球知识图（GKG）数据流汇编而成，覆盖了2013年4月1日至2025年4月14日的时间段，按周（time2id指每周一的日期）聚合实体关系。该数据集的实体来自预先定义的国家和财富500强公司的列表。GDELT-big之所以创建，主要是因为现有的TKG数据集相对稀疏。与其他数据集相比，GDELT-big中的节点在任意随机时间步骤中拥有大于0的度的比例高达22%，远远超过之前可用数据集中的互联程度。该数据集具有广泛的覆盖范围，高密度，以及适合时间序列分析的格式。

GDELT-big is a Temporal Knowledge Graph (TKG) dataset following a format compatible with FinDKG and RE-GCN. It is assembled from the GDELT 1.0 Global Knowledge Graph (GKG) data stream, covering the period from April 1, 2013, to April 14, 2025, with entity relations aggregated per week. The entities are extracted from a pre-defined list of nations and Fortune 500 companies. GDELT-big was created primarily due to the relative sparsity of existing TKG datasets, with a significantly higher interconnectivity among nodes compared to previous datasets. The dataset offers extensive coverage, superior density, and is ready for time-series analysis.