Temporal Neighbourhood Aggregation

Code for the IEEE Big Data 2019 paper entitled 'Temporal Neighbourhood Aggregation: Predicting Future Links in Temporal Graphs via Recurrent Variational Graph Convolutions' A pre-print version of the paper can be found here - https://arxiv.org/abs/1908.08402

This model is designed to predict the next graph snapshot in a temporal graph using recurrent Variational Graph Convolutions.

Dependencies and Requirements

The code has been designed to support python 3.7+ only. The project has the following dependences and version requirements:

• python 3.7+
• pytorch 1.2+
• numpy 1.16+
• scipy 1.1+
• scikit-learn 0.21+
• graph_tool 2.28+

Datasets and Processing

The results in the paper are presented on three empirical graph datasets taken from the Stanford Network Analysis Project:

• Bitcoin-Alpha: https://snap.stanford.edu/data/soc-sign-bitcoin-alpha.html
• Wiki Vote: https://snap.stanford.edu/data/wiki-Vote.html
• UCI Messages: https://snap.stanford.edu/data/index.html

The data_utils.py file can be used to convert the datasets into the numpy matrices needed by the TNA model.

Running the Model

The model can be run using the following command -

python train_TNA.py --dataset='wiki' --seq_len=3 --num_epochs=50 --rnn_model='GRU' --data_loc

Please not that the data_loc flag must be set to the location of your dataset.

Cite

Please cite the associated papers for this work if you use this code:

@inproceedings{bonner2019temporal,
  title={Temporal Neighbourhood Aggregation: Predicting Future Links in Temporal Graphs via Recurrent Variational Graph Convolutions},
  author={Bonner, Stephen and Atapour-Abarghouei, Amir and Jackson, Philip T and Brennan, John and Kureshi, Ibad and Theodoropoulos, Georgios and McGough, Andrew Stephen and Obara, Boguslaw},
  booktitle={2019 IEEE International Conference on Big Data (Big Data)},
  pages={5336--5345},
  year={2019},
  organization={IEEE}
}