TensorFlow and PyTorch implementations of the model proposed in the paper:
Predict then Propagate: Graph Neural Networks meet Personalized PageRank
by Johannes Klicpera, Aleksandar Bojchevski, Stephan Günnemann
Published at ICLR 2019.
The easiest way to get started is by looking at the notebook simple_example_tensorflow.ipynb
or simple_example_pytorch.ipynb
. The notebook reproduce_results.ipynb
shows how to reproduce the results from the paper.
The repository uses these packages:
numpy
scipy
pandas
networkx
tensorflow
pytorch
You can install all requirements via pip install -r requirements.txt
. However, depending on which parts of the repository you use (TensorFlow or PyTorch implementation, NetworkX conversion) you don't need all of these.
To install the package, run python setup.py install
.
In the data
folder you can find several datasets. If you want to use other (external) datasets, you can e.g. use the networkx_to_sparsegraph
method in ppnp.data.io
for converting NetworkX graphs to our SparseGraph format.
The Cora-ML graph was extracted by Aleksandar Bojchevski, and Stephan Günnemann. "Deep gaussian embedding of attributed graphs: Unsupervised inductive learning via ranking." ICLR 2018,
while the raw data was originally published by Andrew Kachites McCallum, Kamal Nigam, Jason Rennie, and Kristie Seymore. "Automating the construction of internet portals with machine learning." Information Retrieval, 3(2):127–163, 2000.
The Citeseer graph was originally published by Prithviraj Sen, Galileo Namata, Mustafa Bilgic, Lise Getoor, Brian Gallagher, and Tina Eliassi-Rad. "Collective Classification in Network Data." AI Magazine, 29(3):93–106, 2008.
The PubMed graph was originally published by Galileo Namata, Ben London, Lise Getoor, and Bert Huang. "Query-driven Active Surveying for Collective Classification". International Workshop on Mining and Learning with Graphs (MLG) 2012.
The Microsoft Academic graph was originally published by Oleksandr Shchur, Maximilian Mumme, Aleksandar Bojchevski, Stephan Günnemann. "Pitfalls of Graph Neural Network Evaluation". Relational Representation Learning Workshop (R2L), NeurIPS 2018.
Please contact [email protected] in case you have any questions.
Please cite our paper if you use the model or this code in your own work:
@inproceedings{klicpera_predict_2019,
title = {Predict then Propagate: Graph Neural Networks meet Personalized PageRank},
author = {Klicpera, Johannes and Bojchevski, Aleksandar and G{\"u}nnemann, Stephan},
booktitle={International Conference on Learning Representations (ICLR)},
year = {2019}
}