What is Learned in Knowledge Graph Embeddings?

TL;DR

This paper investigates what knowledge graph embedding models truly learn, distinguishing between motif learning, network connectivity, and statistical patterns, and evaluates their contributions to model performance.

Contribution

It introduces tests to differentiate the mechanisms behind KG embedding success and applies them to benchmark datasets, challenging assumptions about rule learning.

Findings

KG models can learn motifs but are affected by noise edges

Performance is influenced by multiple mechanisms, not just rule learning

Proposes improved evaluation protocols for KG models

Abstract

A knowledge graph (KG) is a data structure which represents entities and relations as the vertices and edges of a directed graph with edge types. KGs are an important primitive in modern machine learning and artificial intelligence. Embedding-based models, such as the seminal TransE [Bordes et al., 2013] and the recent PairRE [Chao et al., 2020] are among the most popular and successful approaches for representing KGs and inferring missing edges (link completion). Their relative success is often credited in the literature to their ability to learn logical rules between the relations. In this work, we investigate whether learning rules between relations is indeed what drives the performance of embedding-based methods. We define motif learning and two alternative mechanisms, network learning (based only on the connectivity of the KG, ignoring the relation types), and unstructured statistical learning (ignoring the connectivity of the graph). Using experiments on synthetic KGs, we show that KG models can learn motifs and how this ability is degraded by non-motif (noise) edges. We propose tests to distinguish the contributions of the three mechanisms to performance, and apply them to popular KG benchmarks. We also discuss an issue with the standard performance testing protocol and suggest an improvement. To appear in the proceedings of Complex Networks 2021.