Learning Knowledge Graph Embeddings with Type Regularizer

TL;DR

This paper introduces a type regularizer into the RESCAL knowledge graph embedding model, leveraging entity type information to improve relation learning and generalization, with demonstrated performance gains.

Contribution

It proposes a novel type regularizer for RESCAL that incorporates entity type information to enhance knowledge graph embedding quality.

Findings

Improved mean reciprocal rank and hits@N metrics with the regularizer

Type regularizer significantly impacts embedding performance

Scenarios affecting the effectiveness of the type regularizer identified

Abstract

Learning relations based on evidence from knowledge bases relies on processing the available relation instances. Many relations, however, have clear domain and range, which we hypothesize could help learn a better, more generalizing, model. We include such information in the RESCAL model in the form of a regularization factor added to the loss function that takes into account the types (categories) of the entities that appear as arguments to relations in the knowledge base. We note increased performance compared to the baseline model in terms of mean reciprocal rank and hits@N, N = 1, 3, 10. Furthermore, we discover scenarios that significantly impact the effectiveness of the type regularizer.