BiVE: Learning Representations of Bi-level Knowledge Graphs for Reasoning beyond Link Prediction

This code is an implementation of the following paper:

Chanyoung Chung and Joyce Jiyoung Whang, Learning Representations of Bi-level Knowledge Graphs for Reasoning beyond Link Prediction, AAAI Conference on Artificial Intelligence (AAAI), 2023.

This code is based on the OpenKE implementation, which is an open toolkit for knowledge graph embedding. Additional codes are written by Chanyoung Chung (chanyoung.chung@kaist.ac.kr).

When you use this code or data, please cite our paper.

@inproceedings{bive,
	author={Chanyoung Chung and Joyce Jiyoung Whang},
	title={Learning Representations of Bi-level Knowledge Graphs for Reasoning beyond Link Prediction},
	booktitle={Proceedings of the 37th AAAI Conference on Artificial Intelligence},
	year={2023},
	pages={4208--4216},
	doi={10.1609/aaai.v37i4.25538}
}

Usage

Data Augmentation by Random Walks

Use augment.py to perform data augmentation.

python augment.py [data] [conf]

• [data]: name of the dataset. The name should be the directory name of the dataset contained in the ./benchmarks folder.
• [conf]: threshold of the confidence score, i.e., $\tau$ in the paper.

BiVE

To train BiVE-Q, use bive_q_new.py.

CUDA_VISIBLE_DEVICES=0 python bive_q_new.py [data] [learning_rate] [regul_rate] [epoch] --meta [weight_high] --aug [weight_aug] --lp/tp/clp

To train BiVE-B, use bive_b_new.py.

CUDA_VISIBLE_DEVICES=0 python bive_b_new.py [data] [learning_rate] [regul_rate] [epoch] --meta [weight_high] --aug [weight_aug] --lp/tp/clp

License

Our datasets and codes are released under the CC BY-NC-SA 4.0 license.