ToKEi: Temporal Knowledge Graph Embeddings with Arbitrary Time Precision

This is the implementation of the ToKEi models for Temporal Knowledge Graph Embeddings (TKGE) with arbitrary time precision. This code is based on the is original RotatE toolkit.

Requirements

The model is implemented in Python 3, using the PyTorch library. Other library requirements are listed in requirements.txt. To load them into your preference virtual execution environment, use:

pip install -r requirements.txt

Data

Each dataset in stored under a single directory, featuring:

entities.dict # Key values pairs for entities
relations.dict # Key values pairs for relations
temporal # Facts with validity annotations
non-temporal # Optional non-temporal facts.

Usage

Pre-training

    CUDA_VISIBLE_DEVICES=0 python RotatE/run.py                \
            --cuda                                             \
            --do_train                                         \
            --do_valid                                         \
            --do_test                                          \
            --ignore_dict                                      \
            --data_path data/RotatE/wikidata12k                \
            --model RotatE                                     \
            -n 256 -b 1024 -d 1000                             \
            -g 24.0 -a 1.0 -adv                                \
            -lr 0.0001 --max_steps 150000                      \
            -save models/RotatE_WD12k --test_batch_size 16 -de

For a complete list of options, try:

python run.py --help

Training temporal models

    CUDA_VISIBLE_DEVICES=0 python train.py 
            --seed 0 --data_path data/wikidata12k              \
            -n 128 -b 1024 -p 15000 -wu 5000                   \
            -a 1.0 -lr 0.0001 --max_steps 45000                \
            -save models/RotatE_WD12k_CDY --test_batch_size 16 \
            -scope d1100_1_1,d2019_6_30,CDY

For more options, try:

python train.py --help

Testing pre-trained temporal models

    CUDA_VISIBLE_DEVICES=0 python test.py 
            --data_path data/wikidata12k                       \
            -n 128 -b 1024 -p 15000 -wu 5000                   \
            -a 1.0 -lr 0.0001 --max_steps 45000                \
            -save models/RotatE_WD12k_CDY --test_batch_size 16 \
            -scope d1100_1_1,d2019_6_30,CDY                    \
	    --test_levels --test_time --test_scoping --test_ranking

For more options, try:

python test.py --help

Citation

@inproceedings{
  author    = {Julien Leblay and Melisachew Wudage Chekol and Xin Liu},
  title     = {Towards Temporal Knowledge Graph Embeddings with Arbitrary Time Precision},
  booktitle = {Proceedings of the 29th {ACM} International Conference on Information and Knowledge Management ({CIKM} '20), October 19--23, 2020, Virtual Event, Ireland},
  publisher = {{ACM}},
  year      = {2020},
  url       = {https://doi.org/10.1145/3340531.3412028},
  doi       = {10.1145/3340531.3412028},
}

abdul2706 / tokei-semester-project