Resources and code for paper "Fuzzy Logic based Logical Query Answering on Knowledge Graphs".
Make sure your local environment has the following installed:
Python3.9
torch == 1.9.0
wandb == 0.9.7
Install the dependency using:
pip install -r requirements.txt
Download data from here and put it under data folder.
The directory structure should be like [PROJECT_DIR]/data/NELL-betae/train-queries.pkl.
Only FB15k-237 and NELL995 are used in our study.
Training script example: ./run.sh
It usually takes 4 days to a week to finish a run on a NVIDIA® GP102 TITAN Xp (12GB) GPU.
TODO: More training scripts for easy training will be added soon.
The trained model will be automatically stored under the folder ./trained_models. The model name will be [WANDB_RUN_NAME].pt.
To test a trained model, you can use the following command:
python ./test-pretrained-model.py [DATA_NAME] [WANDB_RUN_NAME]
By default, the test tests for product logic. You can also test for other logic systems ('godel' or 'luka') by modifying the logic variable in the script.
The pretrained FuzzQE model (product logic) for NELL can be downloaded here. You can put it under ./trained_models and use the following command to test it:
python ./test-pretrained-model.py NELL feasible-resonance-1518
TODO: More pretrained models will be uploaded soon.
Please refer to our paper if you find the resources useful.
Xuelu Chen, Ziniu Hu, Yizhou Sun. Fuzzy Logic based Logical Query Answering on Knowledge Graphs. Proceedings of the Thirty-sixth AAAI Conference on Artificial Intelligence (AAAI), 2022.
@inproceedings{chen2021fuzzyqa,
title={Fuzzy Logic based Logical Query Answering on Knowledge Graphs},
author={Chen, Xuelu and Hu, Ziniu and Sun, Yizhou}
booktitle={Proceedings of the Thirty-sixth AAAI Conference on Artificial Intelligence (AAAI)},
year={2022}
}