A GitHub (stargazers network) dataset of 1,448,038 nodes and 2,985,781 edges across 12,725 graphs, collected by Rozemberczki et al. via GitHub API (2019). Each record contains graph ID, binary label (0=web development, 1=machine learning), developer-user follower edges. The dataset is notable for its graph classification dataset with 12,725 small graphs, each representing the social network of developers who starred a repository. Intended for research into graph classification, social network analysis, developer behavior modeling, and transfer learning.
A GitHub (user-repository starring/watching network) dataset of 177,386 nodes and 440,237 edges, collected by KONECT via GitHub API. Each record contains directed: False, weighted: False, temporal: False. Bipartite network of GitHub users and repositories — edges represent starring or watching relationships.. Suitable for open source collaboration analysis, bipartite network analysis, software ecosystem research.
A Linux kernel source code (function call graph) dataset of 30,837 nodes and 213,954 edges, collected by KONECT via static analysis of Linux kernel. Each record contains directed: True, weighted: False. Directed call graph of the Linux kernel — nodes are functions, edges are function calls.. Suitable for software network analysis, dependency network research, community detection.
