TabGraphs
收藏资源简介:
TabGraphs是一个包含异构表格节点特征的图数据集基准,由高等经济学院和Yandex研究创建。该数据集包含11个不同的图数据集,涵盖了多种领域和关系类型,旨在评估图神经网络和传统表格模型在处理具有关系信息的表格数据时的性能。数据集的创建过程包括收集和增强现有的表格数据集,通过外部信息构建图结构。TabGraphs的应用领域广泛,旨在解决如何有效利用图结构信息提升表格数据的预测性能问题。
TabGraphs is a benchmark graph dataset with heterogeneous tabular node features, created by researchers from the Higher School of Economics and Yandex Research. This benchmark includes 11 distinct graph datasets covering diverse domains and relationship types, designed to evaluate the performance of graph neural networks and traditional tabular models when handling tabular data with relational information. The development of TabGraphs involves collecting and augmenting existing tabular datasets, and constructing graph structures using external information. With a wide range of application fields, TabGraphs aims to solve the problem of effectively leveraging graph structural information to improve the predictive performance of tabular data.




