CE-GCN

This repo provides a reference implementation of CE-GCN as described in the paper:

Cascade-Enhanced Graph Convolutional Network for Information Diffusion Prediction

Basic Usage

DATASET

You can find the dataset in the "data" folder, which contains all three datasets (Twitter, Douban, and Meme).

Environmental Settings

Our experiments are conducted on CentOS 20.04, a single NVIDIA V100 GPU. CCGL is implemented by Python 3.7, Torch 1.0.9.

Create a virtual environment and install GPU-support packages via Anaconda:

# create virtual environment
conda create --name=CEGCN python=3.9

# activate virtual environment
conda activate CEGCN

# install other related dependencies
conda install wandb

conda install pytorch==1.8.0 torchvision==0.9.0 torchaudio==0.8.0 cudatoolkit=10.2 -c pytorch

conda install pyg -c pyg -c conda-forge

conda install scikit-learn-intelex

Usage

You can run our model with the following commands:

CUDA_VISIBLE_DEVICES=1 python run.py --data="twitter"
CUDA_VISIBLE_DEVICES=1 python run.py --data="douban"
CUDA_VISIBLE_DEVICES=1 python run.py --data="memetracker"

CUDA_VISIBLE_DEVICES=1 nohup python run.py --data="twitter" &
CUDA_VISIBLE_DEVICES=0 nohup python run.py --data="douban" &
CUDA_VISIBLE_DEVICES=1 nohup python run.py --data="memetracker

running with different graphs

CUDA_VISIBLE_DEVICES=1 nohup  python run.py --data="twitter" --notes="item" &
CUDA_VISIBLE_DEVICES=0 nohup  python run.py --data="twitter" --notes="social" &
CUDA_VISIBLE_DEVICES=0 nohup  python run.py --data="twitter" --notes="diffusion" &
CUDA_VISIBLE_DEVICES=1 nohup  python run.py --data="twitter" --notes="social+item" &
CUDA_VISIBLE_DEVICES=1 nohup  python run.py --data="twitter" --notes="diffusion+item" &
CUDA_VISIBLE_DEVICES=1 nohup  python run.py --data="twitter" --notes="social+diffusion" &

CUDA_VISIBLE_DEVICES=1 nohup  python run.py --data="douban" --notes="item" &
CUDA_VISIBLE_DEVICES=0 nohup  python run.py --data="douban" --notes="social" &
CUDA_VISIBLE_DEVICES=1 nohup  python run.py --data="douban" --notes="diffusion" &
CUDA_VISIBLE_DEVICES=0 nohup  python run.py --data="douban" --notes="social+item" &
CUDA_VISIBLE_DEVICES=1 nohup  python run.py --data="douban" --notes="diffusion+item" &
CUDA_VISIBLE_DEVICES=1 nohup  python run.py --data="douban" --notes="social+diffusion" &

Cite

If you find our paper & code are useful for your research, please consider citing us 😘:

@inproceedings{DBLP:conf/dasfaa/WangWYBZZH22,
  author       = {Ding Wang and
                  Lingwei Wei and
                  Chunyuan Yuan and
                  Yinan Bao and
                  Wei Zhou and
                  Xian Zhu and
                  Songlin Hu},
  editor       = {Arnab Bhattacharya and
                  Janice Lee and
                  Mong Li and
                  Divyakant Agrawal and
                  P. Krishna Reddy and
                  Mukesh K. Mohania and
                  Anirban Mondal and
                  Vikram Goyal and
                  Rage Uday Kiran},
  title        = {Cascade-Enhanced Graph Convolutional Network for Information Diffusion
                  Prediction},
  booktitle    = {Database Systems for Advanced Applications - 27th International Conference,
                  {DASFAA} 2022, Virtual Event, April 11-14, 2022, Proceedings, Part
                  {I}},
  series       = {Lecture Notes in Computer Science},
  volume       = {13245},
  pages        = {615--631},
  publisher    = {Springer},
  year         = {2022},
  url          = {https://doi.org/10.1007/978-3-031-00123-9\_50},
  doi          = {10.1007/978-3-031-00123-9\_50},
  timestamp    = {Fri, 29 Apr 2022 14:50:40 +0200},
  biburl       = {https://dblp.org/rec/conf/dasfaa/WangWYBZZH22.bib},
  bibsource    = {dblp computer science bibliography, https://dblp.org}
}

Contact

For any questions please open an issue or drop an email to: [email protected]