WFC
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/zjukg/KGTransformer
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是由三个常见的知识图谱数据集（WN18RR、FB15k-237和Codex）合并而成，这些数据集具有不同的图形属性，以覆盖多种图形结构。其中，WN18RR数据集密集但分布不均，FB15k-237数据集密集且分布均衡，而Codex数据集稀疏且分布不均。该数据集包含了这三个数据集中的多样化图形结构，旨在用于知识图谱的预训练任务。
This dataset is compiled by merging three prevalent knowledge graph datasets: WN18RR, FB15k-237, and Codex. These datasets feature distinct graph properties to cover diverse graph structures. Specifically, WN18RR is dense but unevenly distributed, FB15k-237 is dense and evenly distributed, while Codex is sparse and unevenly distributed. This integrated dataset incorporates the varied graph structures from the three source datasets, and is designed for knowledge graph pre-training tasks.
