Continuous-Time Graph Learning for Cascade Popularity Prediction

This is the implementation of CTCP: [Continuous-Time Graph Learning for Cascade Popularity Prediction, IJCAI 2023].

Requirements

python == 3.7.0
pytorch == 1.9.1
dgl == 0.8.2
scikit-learn == 1.0
numpy == 1.21.5
pandas == 1.1.5
tqdm == 4.62.3

Dataset

Download the preprocessed dataset from Baidu Yun （extract code myu6）
create a directory ./data and put the downloaded dataset into the directory.

Run

Create the directories to store the running results

mkdir log results saved_models

Running command

#Twitter
python main.py --dataset twitter  --prefix std --gpu 0 --epoch 150 --embedding_module aggregate --use_dynamic --use_temporal --use_structural --use_static --dropout 0.6 --predictor merge --lambda 0.1
#APS
python main.py --dataset aps  --prefix std --gpu 0 --epoch 150 --embedding_module aggregate --use_dynamic --use_temporal --use_structural --use_static --dropout 0.6 --predictor merge --lambda 0.1
#Weibo
python main.py --dataset weibo  --prefix std --gpu 0 --epoch 150 --embedding_module aggregate --use_dynamic --use_temporal --use_structural --use_static --dropout 0.6 --predictor merge --lambda 0.1

After running, the log file, results, and trained model are saved under the directories of log, saved_results, and saved_models respectively.

lxd99/CTCP

Continuous-Time Graph Learning for Cascade Popularity Prediction

Requirements

Dataset

Run