🤖 AI Summary
To address the challenges of modeling long-range dependencies, high computational overhead, and oversmoothing in Graph Neural Networks (GNNs) on large-scale heterogeneous graphs, this paper proposes LMSPS—a framework for Long-range Meta-Path Search via Progressive Sampling. Methodologically, LMSPS introduces a novel hop-independent time-complexity progressive random sampling mechanism that dynamically prunes the search space and adaptively selects high-value meta-paths. It is the first approach to enable scalable, jointly optimized meta-path search and importance estimation on heterogeneous graphs. Extensive experiments on multiple large-scale benchmarks demonstrate that LMSPS significantly outperforms state-of-the-art methods: it effectively mitigates oversmoothing, improves search efficiency by over 3×, and achieves optimal performance using fewer than 15% of critical meta-paths.
📝 Abstract
Utilizing long-range dependency, a concept extensively studied in homogeneous graphs, remains underexplored in heterogeneous graphs, especially on large ones, posing two significant challenges: Reducing computational costs while maximizing effective information utilization in the presence of heterogeneity, and overcoming the over-smoothing issue in graph neural networks. To address this gap, we investigate the importance of different meta-paths and introduce an automatic framework for utilizing long-range dependency on heterogeneous graphs, denoted as Long-range Meta-path Search through Progressive Sampling (LMSPS). Specifically, we develop a search space with all meta-paths related to the target node type. By employing a progressive sampling algorithm, LMSPS dynamically shrinks the search space with hop-independent time complexity. Through a sampling evaluation strategy, LMSPS conducts a specialized and effective meta-path selection, leading to retraining with only effective meta-paths, thus mitigating costs and over-smoothing. Extensive experiments across diverse heterogeneous datasets validate LMSPS's capability in discovering effective long-range meta-paths, surpassing state-of-the-art methods. Our code is available at https://github.com/JHL-HUST/LMSPS.