HyperKG: Hyperbolic Knowledge Graph Embeddings for Knowledge Base Completion

TL;DR

HyperKG introduces hyperbolic space embeddings for knowledge bases, significantly improving the performance of translational models in link prediction tasks by better capturing the data's topological properties.

Contribution

The paper proposes HyperKG, a novel hyperbolic embedding model that enhances translational models for knowledge base completion, capturing complex topological regularities.

Findings

HyperKG narrows the performance gap between translational and bilinear models.

Hyperbolic space better reflects the topological properties of knowledge bases.

The model effectively captures a subset of Datalog rules.

Abstract

Learning embeddings of entities and relations existing in knowledge bases allows the discovery of hidden patterns in data. In this work, we examine the geometrical space's contribution to the task of knowledge base completion. We focus on the family of translational models, whose performance has been lagging, and propose a model, dubbed HyperKG, which exploits the hyperbolic space in order to better reflect the topological properties of knowledge bases. We investigate the type of regularities that our model can capture and we show that it is a prominent candidate for effectively representing a subset of Datalog rules. We empirically show, using a variety of link prediction datasets, that hyperbolic space allows to narrow down significantly the performance gap between translational and bilinear models.