Link Prediction with Relational Hypergraphs

TL;DR

This paper introduces a new framework for link prediction in relational hypergraphs, extending graph neural network techniques to handle complex $k$-ary relations, with theoretical analysis and empirical validation showing superior performance.

Contribution

The paper presents a novel framework for link prediction in relational hypergraphs, including theoretical analysis of expressive power and state-of-the-art empirical results.

Findings

Models outperform all baselines in inductive link prediction

Achieve state-of-the-art results in transductive link prediction

Theoretical analysis confirms expressive capabilities of the models

Abstract

Link prediction with knowledge graphs has been thoroughly studied in graph machine learning, leading to a rich landscape of graph neural network architectures with successful applications. Nonetheless, it remains challenging to transfer the success of these architectures to relational hypergraphs, where the task of link prediction is over $k$-ary relations, which is substantially harder than link prediction with knowledge graphs. In this paper, we propose a framework for link prediction with relational hypergraphs, unlocking applications of graph neural networks to fully relational structures. Theoretically, we conduct a thorough analysis of the expressive power of the resulting model architectures via corresponding relational Weisfeiler-Leman algorithms and also via logical expressiveness. Empirically, we validate the power of the proposed model architectures on various relational hypergraph benchmarks. The resulting model architectures substantially outperform every baseline for inductive link prediction, and lead to state-of-the-art results for transductive link prediction.