WIKI-CS

WIKI-CS是由剑桥大学计算机科学系创建的一个基于维基百科的数据集，用于评估图神经网络。该数据集包含11701个节点，对应计算机科学文章，边基于超链接，分为10个类别，代表该领域的不同分支。数据集的创建过程涉及从维基百科数据中提取子图，并通过特定的分类工具进行清理和分类。WIKI-CS主要用于半监督节点分类和单关系链接预测，旨在解决现有基准数据集在结构多样性和评估标准上的不足。WIKI-CS is a Wikipedia-based dataset created by the Department of Computer Science at the University of Cambridge for evaluating graph neural networks. This dataset contains 11,701 nodes corresponding to computer science articles, with edges constructed based on hyperlinks, and is divided into 10 categories representing different subfields of computer science. The dataset creation process involves extracting subgraphs from Wikipedia data, followed by cleaning and classification using specialized classification tools. WIKI-CS is primarily used for semi-supervised node classification and single-relation link prediction, aiming to address the shortcomings of existing benchmark datasets in terms of structural diversity and evaluation criteria.

搜集汇总

背景与挑战

背景概述

Wiki-CS是一个基于维基百科的图神经网络基准测试数据集，包含节点特征、邻接列表和标签，并提供了20个不同的训练分割以减少过拟合风险。它专为节点分类和链接预测任务设计，支持通过PyTorch Geometric等工具加载，适用于评估图神经网络模型的性能。