Knowledge graph autoencoder training environment with modular design and reproducibility
KGATE : a Knowledge Graph Embedding Training Environment
Machine LearningArtificial Intelligence
Summary
Knowledge graphs store information as entities connected by relationships. To better use these graphs for tasks like predicting missing links or classifying data, they can be converted into a simpler form called embeddings. The authors present KGATE, a software library that helps train these embeddings using an approach called an autoencoder, which compresses and then reconstructs the graph data. KGATE is designed to be easy to customize, prevent data leaks, and ensure results can be reliably reproduced. It performs training as fast as the best existing tools but offers more flexibility and features.
What this means in practice
- •For machine learning engineers: Build and train knowledge graph embedding models with customizable components and reproducible pipelines for enhanced experimentation.
- •For data integration teams: Use KGATE to preprocess and embed knowledge graph data while controlling for data leakage to improve downstream analytics.
Authors
Benjamin Loire, Galadriel Brière, Célia Brahimi, Antoine Toffano, Anaïs Baudot
Abstract
Knowledge graph embedding (KGE) models encode the entities and relations of a knowledge graph into a low-dimensional latent space, enabling tasks such as classification or link prediction. Most KGE models follow an autoencoder architecture, in which an encoder projects the knowledge graph into the latent space and a decoder reconstruct it. Combining both encoder and decoder components is increasingly needed, yet existing libraries rarely support complete autoencoders, are often unmaintained, rely on undocumented default hyperparameters, and produce results that cannot be compared across libraries. Here we present KGATE (Knowledge Graph Autoencoder Training Environment), a modular Python library built on PyTorch Geometric and TorchKGE. KGATE lets users assemble initializers, encoders, decoders, losses, negative samplers, and evaluation metrics as building blocks, or plug in their own block. KGATE includes a preprocessing procedure that controls data leakage, a builtin training pipeline, and reproducibility by design. Benchmarks against six existing KGE libraries show that KGATE training time is comparable with the fastest libraries while offering a broader set of features.