Explore high-quality datasets for your AI and machine learning projects.
This dataset is designed for one‑to‑many graph translation tasks. The graphs have no node features; the goal is to learn a mapping from input graph topology to target graph topology. Each input graph is a directed scale‑free network with a power‑law degree distribution. For target graph generation, a node is selected as the target node with probability proportional to its indegree and connected to a new source node with probability 0.41. Similarly, a source node is chosen proportional to its outdegree and connects to a new target node with probability 0.54. Then, m edges (where m equals the number of input nodes) are added between the two nodes to form the target graph. Hence both input and target graphs are directed scale‑free graphs.
