A First Experiment on Including Text Literals in KGloVe

TL;DR

This paper explores incorporating textual literal properties into graph embedding models to enhance knowledge graph representations, presenting initial experimental results that suggest potential but no clear performance improvement.

Contribution

It introduces a novel approach combining graph structure with literal textual information in embeddings, an area not extensively explored before.

Findings

Initial experiments show potential benefits of including text literals.

The approach does not outperform existing methods on current ML tasks.

Further research is needed to optimize the integration of textual literals.

Abstract

Graph embedding models produce embedding vectors for entities and relations in Knowledge Graphs, often without taking literal properties into account. We show an initial idea based on the combination of global graph structure with additional information provided by textual information in properties. Our initial experiment shows that this approach might be useful, but does not clearly outperform earlier approaches when evaluated on machine learning tasks.