GraphGAN

In this repository, We implement GraphGAN (arXiv) using tensorflow.

GraphGAN: Graph Representation Learning With Generative Adversarial Nets
Hongwei Wang, Jia Wang, Jialin Wang, Miao Zhao, Weinan Zhang, Fuzheng Zhang, Xing Xie, Minyi Guo
32nd AAAI Conference on Artificial Intelligence, 2018

Requirements

The code has been tested running under Python 3.6.5, with the following packages installed (along with their dependencies):

• tensorflow == 1.8.0
• tqdm == 4.23.4
• numpy == 1.14.3
• sklearn == 0.19.1

Files in the folder

• data/: the preprocessed data and the pre-trained embedding for each experiment
• exp: the scripts used to perform experiments
• graphGAN/: source codes

How to use this repository?

For simplicity, we use CODE_DIR to denote the path that you store this repository.

Data preprocess

We take the recommendation experiment using dataset MovieLens-1M for example. The edges are stored in ${CODE_DIR}/data/ml-1m/all_edges.txt with the following format:

0 1
3 2
...

cd ${CODE_DIR}/data/ml-1m
python ${CODE_DIR}/graphGAN/utils/recommendation.py all_edges.txt

This step will divide the training set and the test set and also generate train_trees.pkl used in the subsequent training step.

Training

cd ${CODE_DIR}/exp/ml-1m
export PYTHONPATH=/data/private/ws/projects/GraphGAN/ShuoGraphGAN:$PYTHONPATH
mkdir -p ./train
python ${CODE_DIR}/bin/trainer.py \
--data_dir ${CODE_DIR}/data/ml-1m \
--log_dir ./train

The trained models and evaluation results are stored in ./train. You can also evaluate the trained model with scorer.py:

python ${CODE_DIR}/bin/scorer.py \
--data_dir ${CODE_DIR}/data/ml-1m \
--log_dir ./train