Awesome

Installation

The latest stable version of PyKEEN requires Python 3.9+. It can be downloaded and installed from PyPI with:

pip install pykeen

The latest version of PyKEEN can be installed directly from the source code on GitHub with:

pip install git+https://github.com/pykeen/pykeen.git

More information about installation (e.g., development mode, Windows installation, Colab, Kaggle, extras) can be found in the installation documentation.

Quickstart

This example shows how to train a model on a dataset and test on another dataset.

The fastest way to get up and running is to use the pipeline function. It provides a high-level entry into the extensible functionality of this package. The following example shows how to train and evaluate the TransE model on the Nations dataset. By default, the training loop uses the stochastic local closed world assumption (sLCWA) training approach and evaluates with rank-based evaluation.

from pykeen.pipeline import pipeline

result = pipeline(
    model='TransE',
    dataset='nations',
)

The results are returned in an instance of the PipelineResult dataclass that has attributes for the trained model, the training loop, the evaluation, and more. See the tutorials on using your own dataset, understanding the evaluation, and making novel link predictions.

PyKEEN is extensible such that:

• Each model has the same API, so anything from pykeen.models can be dropped in
• Each training loop has the same API, so pykeen.training.LCWATrainingLoop can be dropped in
• Triples factories can be generated by the user with from pykeen.triples.TriplesFactory

The full documentation can be found at https://pykeen.readthedocs.io.

Implementation

Below are the models, datasets, training modes, evaluators, and metrics implemented in pykeen.

Datasets

The following 37 datasets are built in to PyKEEN. The citation for each dataset corresponds to either the paper describing the dataset, the first paper published using the dataset with knowledge graph embedding models, or the URL for the dataset if neither of the first two are available. If you want to use a custom dataset, see the Bring Your Own Dataset tutorial. If you have a suggestion for another dataset to include in PyKEEN, please let us know here.

Inductive Datasets

The following 5 inductive datasets are built in to PyKEEN.

Representations

The following 20 representations are implemented by PyKEEN.

Interactions

The following 34 interactions are implemented by PyKEEN.

Models

The following 40 models are implemented by PyKEEN.

Losses

The following 15 losses are implemented by PyKEEN.

Regularizers

The following 6 regularizers are implemented by PyKEEN.

Training Loops

The following 3 training loops are implemented in PyKEEN.

Negative Samplers

The following 3 negative samplers are implemented in PyKEEN.

Stoppers

The following 2 stoppers are implemented in PyKEEN.

Evaluators

The following 5 evaluators are implemented in PyKEEN.

Metrics

The following 44 metrics are implemented in PyKEEN.

Trackers

The following 8 trackers are implemented in PyKEEN.

Experimentation

Reproduction

PyKEEN includes a set of curated experimental settings for reproducing past landmark experiments. They can be accessed and run like:

pykeen experiments reproduce tucker balazevic2019 fb15k

Where the three arguments are the model name, the reference, and the dataset. The output directory can be optionally set with -d.

Ablation

PyKEEN includes the ability to specify ablation studies using the hyper-parameter optimization module. They can be run like:

pykeen experiments ablation ~/path/to/config.json

Large-scale Reproducibility and Benchmarking Study

We used PyKEEN to perform a large-scale reproducibility and benchmarking study which are described in our article:

@article{ali2020benchmarking,
  author={Ali, Mehdi and Berrendorf, Max and Hoyt, Charles Tapley and Vermue, Laurent and Galkin, Mikhail and Sharifzadeh, Sahand and Fischer, Asja and Tresp, Volker and Lehmann, Jens},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
  title={Bringing Light Into the Dark: A Large-scale Evaluation of Knowledge Graph Embedding Models under a Unified Framework},
  year={2021},
  pages={1-1},
  doi={10.1109/TPAMI.2021.3124805}}
}

We have made all code, experimental configurations, results, and analyses that lead to our interpretations available at https://github.com/pykeen/benchmarking.

Contributing

Contributions, whether filing an issue, making a pull request, or forking, are appreciated. See CONTRIBUTING.md for more information on getting involved.

If you have questions, please use the GitHub discussions feature at https://github.com/pykeen/pykeen/discussions/new.

Acknowledgements

Supporters

This project has been supported by several organizations (in alphabetical order):

• Bayer
• CoronaWhy
• Enveda Biosciences
• Fraunhofer Institute for Algorithms and Scientific Computing
• Fraunhofer Institute for Intelligent Analysis and Information Systems
• Fraunhofer Center for Machine Learning
• Harvard Program in Therapeutic Science - Laboratory of Systems Pharmacology
• Ludwig-Maximilians-Universität München
• Munich Center for Machine Learning (MCML)
• Siemens
• Smart Data Analytics Research Group (University of Bonn & Fraunhofer IAIS)
• Technical University of Denmark - DTU Compute - Section for Cognitive Systems
• Technical University of Denmark - DTU Compute - Section for Statistics and Data Analysis
• University of Bonn

Funding

The development of PyKEEN has been funded by the following grants:

Logo

The PyKEEN logo was designed by Carina Steinborn

Citation

If you have found PyKEEN useful in your work, please consider citing our article:

@article{ali2021pykeen,
    author = {Ali, Mehdi and Berrendorf, Max and Hoyt, Charles Tapley and Vermue, Laurent and Sharifzadeh, Sahand and Tresp, Volker and Lehmann, Jens},
    journal = {Journal of Machine Learning Research},
    number = {82},
    pages = {1--6},
    title = {{PyKEEN 1.0: A Python Library for Training and Evaluating Knowledge Graph Embeddings}},
    url = {http://jmlr.org/papers/v22/20-825.html},
    volume = {22},
    year = {2021}
}