The GNN models and the dataset used in the DRGN
Here we provide papers that generate or analyze the four GNN models respectively. If you want to know the specific details of the three GNN models, please read the following papers:
- PageRank: Graph neural networks for ranking Web pages
- GCN for node classification: Semi-Supervised Classification with Graph Convolutional Networks
- GCN for graph classification: How Powerful are Graph Neural Networks?
- GraphSage: Inductive representation learning on large graphs
This directory contains code necessary to run the different GNN models.
Note: The code of the GNN models here are quoted from the open source code provided by other authors. Here we list the URL of their open source code repository, and more details can be obtained from it.
- gcn_n:tkipf/gcn: Implementation of Graph Convolutional Networks in TensorFlow (github.com)
- gcn_g:weihua916/powerful-gnns: How Powerful are Graph Neural Networks? (github.com)
- graphsage:williamleif/GraphSAGE: Representation learning on large graphs using stochastic graph convolutions. (github.com)
The PageRank dataset used in the paper is a small dataset generated form the DGL tutorials. Here we provide the code to generate this dataset using the DGL library
PageRank dataset
Node: 100
Edge: 629