Directed Homophily-Aware Graph Neural Network

2025-05-28

AI Summary

Existing graph neural networks (GNNs) exhibit limited generalization on directed graphs, struggling to jointly model directional dependencies and neighborhood heterogeneity. To address this, we propose the first GNN framework that simultaneously incorporates direction awareness and homophily adaptivity. Our method introduces a resettable gating mechanism to dynamically modulate message contributions and a structure-aware, noise-robust bidirectional fusion module—marking the first explicit modeling of directional homophily disparities and inter-layer homophily fluctuations. The architecture integrates reverse-directional graph representations, homophily-adaptive aggregation, and robust feature fusion. Extensive experiments on diverse directed homophilic and heterophilic graph benchmarks demonstrate significant improvements in node classification and link prediction. Notably, our approach achieves up to 15.07% absolute gain over state-of-the-art baselines in link prediction.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Graph Neural Networks (GNNs) have achieved significant success in various learning tasks on graph-structured data. Nevertheless, most GNNs struggle to generalize to heterophilic neighborhoods. Additionally, many GNNs ignore the directional nature of real-world graphs, resulting in suboptimal performance on directed graphs with asymmetric structures. In this work, we propose Directed Homophily-aware Graph Neural Network (DHGNN), a novel framework that addresses these limitations by incorporating homophily-aware and direction-sensitive components. DHGNN employs a resettable gating mechanism to adaptively modulate message contributions based on homophily levels and informativeness, and a structure-aware noise-tolerant fusion module to effectively integrate node representations from the original and reverse directions. Extensive experiments on both homophilic and heterophilic directed graph datasets demonstrate that DHGNN outperforms state-of-the-art methods in node classification and link prediction. In particular, DHGNN improves over the best baseline by up to 15.07% in link prediction. Our analysis further shows that the gating mechanism captures directional homophily gaps and fluctuating homophily across layers, providing deeper insights into message-passing behavior on complex graph structures.

Problem

• Improving GNN performance in heterophilic neighborhoods
• Addressing directional nature neglect in real-world graphs
• Enhancing node classification and link prediction accuracy

Innovation

• Homophily-aware gating for adaptive message modulation
• Direction-sensitive fusion for asymmetric graph structures
• Noise-tolerant module integrating reverse-direction representations