Adaptive Heterogeneous Graph Neural Networks: Bridging Heterophily and Heterogeneity

TL;DR

This paper introduces AHGNN, a novel graph neural network designed to effectively handle heterophily and heterogeneity in real-world heterogeneous graphs, improving performance where traditional methods struggle.

Contribution

The paper proposes a heterophily-aware convolution and a coarse-to-fine attention mechanism to model heterophilic heterogeneous graphs more accurately.

Findings

AHGNN outperforms 20 baselines on 7 real-world graphs.

It achieves superior results in high-heterophily scenarios.

The method effectively captures diverse semantic information across meta-paths.

Abstract

Heterogeneous graphs (HGs) are common in real-world scenarios and often exhibit heterophily. However, most existing studies focus on either heterogeneity or heterophily in isolation, overlooking the prevalence of heterophilic HGs in practical applications. Such ignorance leads to their performance degradation. In this work, we first identify two main challenges in modeling heterophily HGs: (1) varying heterophily distributions across hops and meta-paths; (2) the intricate and often heterophily-driven diversity of semantic information across different meta-paths. Then, we propose the Adaptive Heterogeneous Graph Neural Network (AHGNN) to tackle these challenges. AHGNN employs a heterophily-aware convolution that accounts for heterophily distributions specific to both hops and meta-paths. It then integrates messages from diverse semantic spaces using a coarse-to-fine attention mechanism, which filters out noise and emphasizes informative signals. Experiments on seven real-world graphs and twenty baselines demonstrate the superior performance of AHGNN, particularly in high-heterophily situations.