On the Effectiveness of Knowledge Graph Embeddings: a Rule Mining Approach

TL;DR

This paper investigates how different knowledge graph embedding methods affect rule mining outcomes for KG completion, revealing significant differences and potential spurious rules introduced by certain approaches.

Contribution

It introduces a comparative analysis of rule extraction before and after KG completion using various embedding models, highlighting the impact on rule quality.

Findings

Different KGE methods produce significantly different rules.

TransE completion can lead to spurious rule extraction.

Rule quality varies depending on the embedding approach used.

Abstract

We study the effectiveness of Knowledge Graph Embeddings (KGE) for knowledge graph (KG) completion with rule mining. More specifically, we mine rules from KGs before and after they have been completed by a KGE to compare possible differences in the rules extracted. We apply this method to classical KGEs approaches, in particular, TransE, DistMult and ComplEx. Our experiments indicate that there can be huge differences between the extracted rules, depending on the KGE approach for KG completion. In particular, after the TransE completion, several spurious rules were extracted.