UniKG
收藏资源简介:
UniKG是由南京理工大学构建的大型异构图基准数据集,源自Wikidata,旨在促进知识挖掘和异构图表示学习。该数据集包含超过7700万具有多属性的实体和2000种多样化的关联类型,显著超越现有异构图数据集的规模。创建过程中,采用了语义对齐策略和创新的插件式各向异性传播模块(APM),以实现大规模异构图中的高效信息传播和多属性关联的自适应挖掘。UniKG的应用领域广泛,旨在解决大规模异构图学习和知识提取的问题,特别是在推荐系统、恶意软件检测系统和医疗健康系统等领域。
UniKG is a large heterogeneous graph benchmark dataset constructed by Nanjing University of Science and Technology, sourced from Wikidata, aiming to facilitate knowledge mining and heterogeneous graph representation learning. This dataset contains over 77 million entities with multiple attributes and 2000 diverse association types, significantly exceeding the scale of existing heterogeneous graph datasets. During its development, semantic alignment strategies and an innovative pluggable anisotropic propagation module (APM) were adopted to achieve efficient information propagation in large-scale heterogeneous graphs and adaptive mining of multi-attribute associations. UniKG has a wide range of application scenarios, aiming to address the challenges of large-scale heterogeneous graph learning and knowledge extraction, particularly in fields such as recommendation systems, malware detection systems, and healthcare systems.

- 1UniKG: A Benchmark and Universal Embedding for Large-Scale Knowledge Graphs南京理工大学 · 2023年



