LogicENN: A Neural Based Knowledge Graphs Embedding Model with Logical Rules

TL;DR

LogicENN is a neural knowledge graph embedding model that effectively incorporates logical rules, including complex relations, and outperforms existing models in link prediction tasks.

Contribution

It introduces a novel neural embedding framework capable of learning and integrating various logical rules without grounding, a first in neural models.

Findings

LogicENN outperforms state-of-the-art models in link prediction.

The model can learn all ground truths of encoded rules.

It effectively incorporates multiple types of logical rules.

Abstract

Knowledge graph embedding models have gained significant attention in AI research. Recent works have shown that the inclusion of background knowledge, such as logical rules, can improve the performance of embeddings in downstream machine learning tasks. However, so far, most existing models do not allow the inclusion of rules. We address the challenge of including rules and present a new neural based embedding model (LogicENN). We prove that LogicENN can learn every ground truth of encoded rules in a knowledge graph. To the best of our knowledge, this has not been proved so far for the neural based family of embedding models. Moreover, we derive formulae for the inclusion of various rules, including (anti-)symmetric, inverse, irreflexive and transitive, implication, composition, equivalence and negation. Our formulation allows to avoid grounding for implication and equivalence relations. Our experiments show that LogicENN outperforms the state-of-the-art models in link prediction.