Complex Networks

Network datasets are lists of nodes and edges, used to test graph algorithms, community detection and link prediction. The big collections in this list are the Stanford Large Network Dataset Collection, the Koblenz Network Collection, Network Repository, the UCI Network Data Repository and the Laboratory for Web Algorithmics at UNIMI, which publishes large web and social graphs. The UFL sparse matrix collection offers many graphs as sparse matrices.

Domain-specific graphs include AMiner and DBLP citation data, NBER patent citations, DIMACS road networks, a PyPI and Maven dependency network, Stanford GraphBase and a Steam co-review network. Check whether edges are directed, weighted or timestamped, and which file format is used: plain edge lists, Matrix Market and the compressed WebGraph format each need different loaders. The largest graphs will not fit in memory with general-purpose libraries like NetworkX.