PLEASING

Temporal Knowledge Graph Reasoning.

PLEASING: Exploring the Historical and Potential Events for Temporal Knowledge Graph Reasoning(Under Revise)

Dependencies

The required framework and other libraries can be found in the requirements.txt.

Note that GDELT preprocessing requires more memory.

Commands

Preprocess:

cd data/dataset
python get_history_graph.py

Training and Testing:

All hyper-parameter settings can be found in our paper.

YAGO

python main_tt.py -d YAGO --description yago_hard --max-epochs 30 --oracle-epochs 20 --valid-epochs 5 --alpha 0.1 --lambdax 2 --batch-size 1024 --lr 0.001 --oracle_lr 0.001 --oracle_mode hard --save_dir SAVE --eva_dir SAVE --k 15 --beta 0.6 --dropout 0.2 --gamma 0.1 --static False --time_span 1 --timestamps 189

ICEWS18

python main_tt.py -d ICEWS18 --description icews18_soft --max-epochs 50 --oracle-epochs 20 --valid-epochs 10 --alpha 0.2 --lambdax 2 --batch-size 1024 --lr 0.001 --oracle_lr 0.001 --oracle_mode soft --save_dir SAVE --eva_dir SAVE --k 45 --beta 0.6 --gamma 0.1 --dropout 0.2

ICEWS14

python main_tt.py -d ICEWS14T --description icews14T_soft --max-epochs 50 --oracle-epochs 20 --valid-epochs 10 --alpha 0.2 --lambdax 2 --batch-size 1024 --lr 0.001 --oracle_lr 0.001 --oracle_mode soft --save_dir SAVE --eva_dir SAVE --k 45 --beta 0.6 --gamma 0.1 --dropout 0.2 --static False --time_span 24 --timestamps 365

GDELT

python main_tt.py -d GDELT --description gdelt_soft --max-epochs 30 --oracle-epochs 20 --valid-epochs 10 --alpha 0.2 --lambdax 2 --batch-size 1024 --lr 0.001 --oracle_lr 0.001 --oracle_mode soft --save_dir SAVE --eva_dir SAVE --k 15 --beta 0.6 --gamma 0.1 --dropout 0.2 --time_span 15 --timestamps 2976 --static False

Acknowledge

Some of our code is also referenced from CENET, and the original dataset can be found here: https://github.com/xyjigsaw/CENET.

And RE-GCN: https://github.com/Lee-zix/RE-GCN

Citation

@article{zhang-etal-2023-pleasing,
title = {PLEASING: Exploring the Historical and Potential Events for Temporal Knowledge Graph Reasoning},
journal = {Neural Networks},
year = {2023},
author = {Jinchuan Zhang, Ming Sun, Qian Huang, Ling Tian},
keywords = {Temporal knowledge graphs; Extrapolation; Representation learning; Contrastive learning}
}