Introduction to graphs data.

XFlow incorporates existing dataset collections and can be extended easily. Specifically, our implementation is compatible with any graph in NetworkX 1 graph object, including graphs such as Watts-Strogatz small-world graphs, Barabási–Albert (BA), Erdős–Rényi (ER). See the full list in Graph Generators in NetworkX; Or graph objects in PyTorch Geometric such as Cora, Cite Seer, PubMed and co-purchasing networks (Amazon Photo, Computers) as well as synthetic graph generators. See torch_geometric.datasets in PyTorch Geometric 2.

Networkx graph objects are collected in xflow.dataset.nx package, and the user can import them by the code below (BA graph as example)

from  xflow.dataset.nx import BA

PyG graph obejects are collected in xflow.dataset.pyg package, and the user can use one PyG graph by (Cora as example)

from  xflow.dataset.pyg import cora

  1. NetworkX ↩︎

  2. PyG ↩︎