遇见数据集

stableGCN_datasets

收藏
DataCite Commons2024-09-05 更新2025-04-16 收录
官方服务:

资源简介:

Cora, Citeseer, and Pubmed are commonly used citation network datasets. Among these, Citeseer has the most dense features, while Pubmed has more nodes and edges.ACM is a network of papers where each node represents a paper. In contrast to citation networks, edges connect papers that share the same authors.Flickr serves as a social network that captures the connections between users originating from image and video hosting websites. These users are categorized into nine groups according to their personal interests.UAI is a popular dataset used to evaluate GCNs performance on community detection tasks. It comprises a webpage citation network where nodes represent webpages gathered from various universities, with each edge indicating a citation.Amazon-P and Amazon-C are segments of the Amazon co-purchase graph. In this graph, nodes represent products, edges denote frequent co-purchases between two products, node features are encoded product reviews using bag-of-words, and class labels correspond to product categories.

Cora、Citeseer与Pubmed均为常用的引用网络数据集。其中,Citeseer的特征密度最高,而Pubmed包含的节点与边数量更多。ACM是一种论文关联网络,其中每个节点代表一篇论文;与引用网络不同的是,其边连接的是拥有共同作者的论文。Flickr是一款社交网络数据集,用于刻画源自图像与视频托管网站的用户之间的关联关系,这些用户会根据个人兴趣被划分为九个组别。UAI是一款常用于评估图卷积神经网络(Graph Convolutional Network, GCN)在社区检测任务中性能的数据集,它由网页引用网络构成,节点代表从多所高校采集的网页,每条边均表示一条引用关系。Amazon-P与Amazon-C是亚马逊共同购买图的两个子集,在该图中,节点代表商品,边表示两件商品间存在频繁的共同购买行为,节点特征为使用词袋模型(bag-of-words)对商品评论进行编码得到的特征,类别标签则对应商品的分类。

提供机构:
IEEE DataPort
创建时间:
2024-09-05
搜集汇总
背景与挑战
背景概述
stableGCN_datasets是一个包含多个图网络数据集的集合,涵盖引文网络(如Cora、Citeseer、Pubmed)、论文网络(ACM)、社交网络(Flickr)、网页引文网络(UAI)和产品共购图(Amazon-P、Amazon-C)。这些数据集具有不同的节点类型、边连接方式和特征密度,常用于图卷积网络(GCN)的性能评估,适用于社区检测、分类等任务,体现了多领域应用和结构多样性。
以上内容由遇见数据集搜集并总结生成
二维码
社区交流群
二维码
科研交流群
商业服务