On The Expressive Power of Knowledge Graph Embedding Methods

TL;DR

This paper introduces a mathematical framework to compare the reasoning capabilities of knowledge graph embedding methods, demonstrating that STransE outperforms TransComplEx and proposing an improved STransCoRe method.

Contribution

It provides a novel framework for analyzing KGE reasoning abilities and introduces an improved embedding method, STransCoRe, that reduces complexity while enhancing reasoning.

Findings

STransE has higher reasoning capability than TransComplEx.

The STransCoRe method improves STransE by combining it with TransCoRe insights.

The new method reduces space complexity of embeddings.

Abstract

Knowledge Graph Embedding (KGE) is a popular approach, which aims to represent entities and relations of a knowledge graph in latent spaces. Their representations are known as embeddings. To measure the plausibility of triplets, score functions are defined over embedding spaces. Despite wide dissemination of KGE in various tasks, KGE methods have limitations in reasoning abilities. In this paper we propose a mathematical framework to compare reasoning abilities of KGE methods. We show that STransE has a higher capability than TransComplEx, and then present new STransCoRe method, which improves the STransE by combining it with the TransCoRe insights, which can reduce the STransE space complexity.