WD27M, FB37M
收藏arXiv2023-12-18 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/2312.11053v1
下载链接
链接失效反馈官方服务：
资源简介：
本研究构建了两个新的基准数据集WD27M和FB37M，用于时间知识图谱的冲突检测。WD27M基于Wikidata，包含2700万事实，而FB37M基于Freebase，包含3700万事实。这两个数据集通过半自动方法注释，用于评估自动约束挖掘方法的有效性。数据集的应用领域是时间知识图谱的质量管理和冲突检测，旨在解决时间一致性维护中的新挑战。
This study develops two novel benchmark datasets, WD27M and FB37M, for conflict detection in temporal knowledge graphs. WD27M is derived from Wikidata and contains 27 million facts, while FB37M is based on Freebase and includes 37 million facts. Both datasets are annotated via a semi-automatic approach and are utilized to evaluate the effectiveness of automated constraint mining methods. The application scenarios of these datasets cover quality management and conflict detection for temporal knowledge graphs, aiming to address emerging challenges in temporal consistency maintenance.
提供机构：
南京大学软件新技术国家重点实验室
创建时间：
2023-12-18
