pykeen
🤖 A Python library for learning and evaluating knowledge graph embeddings
What is pykeen?
🤖 A Python library for learning and evaluating knowledge graph embeddings
Overview of pykeen
What it is
PyKEEN (Python KnowlEdge EmbeddiNgs) is a Python library for learning and evaluating knowledge graph embedding models. It supports training and evaluation of models that incorporate multi-modal information.
Key features
- Provides a high-level pipeline function for training and evaluating models with minimal code - Includes 37 built-in datasets (e.g., FB15k-237, BioKG, Hetionet) and 5 inductive datasets - Implements over 40 knowledge graph embedding models with a consistent API - Supports multiple training modes, including the stochastic local closed world assumption (sLCWA)
- Offers rank-based evaluation and extensible components for custom models, training loops, and triples factories
Use cases
- Training and evaluating knowledge graph embedding models for link prediction - Reproducing benchmark results on standard datasets - Developing and testing new embedding models or training strategies - Integrating knowledge graph embeddings into downstream applications such as drug repurposing or question answering
Getting started
Install the latest stable version from PyPI: shell pip install pykeen Or install the latest development version from GitHub: shell pip install git+https://github.com/pykeen/pykeen.git A minimal training example: python from pykeen.pipeline import pipeline result = pipeline( model='TransE', dataset='nations', )
License and pricing
The project is licensed under the MIT License. It is open source and free to use.
Based on the project's public README and repository metadata.