GNNs Meet Sequence Models Along the Shortest-Path: an Expressive Method for Link Prediction

2025-07-08

AI Summary

Graph neural networks (GNNs) struggle to capture pair-specific subgraph structures in link prediction, as their message-passing mechanisms inherently ignore topological details of connecting paths; existing structural enhancement approaches either incur prohibitive computational overhead or rely solely on shallow heuristics (e.g., common neighbors), failing to model multi-hop dependencies. This paper proposes PathGNN: the first GNN-based link prediction framework that explicitly incorporates shortest-path sequence modeling—first generating node embeddings via a base GNN, then enumerating all shortest paths between candidate node pairs, and finally aggregating node embeddings along each path using a sequence model (e.g., Transformer). We theoretically prove that PathGNN’s expressive power strictly surpasses both standard GNNs and mainstream structural feature methods. Empirically, PathGNN achieves state-of-the-art performance across multiple benchmark datasets, while maintaining linear time complexity and high representational capacity.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Search and Optimization: Learning to Search

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
• Social Networks and Social Media: Influence propagation, information diffusion, and the prediction on networks

Abstract

Graph Neural Networks (GNNs) often struggle to capture the link-specific structural patterns crucial for accurate link prediction, as their node-centric message-passing schemes overlook the subgraph structures connecting a pair of nodes. Existing methods to inject such structural context either incur high computational cost or rely on simplistic heuristics (e.g., common neighbor counts) that fail to model multi-hop dependencies. We introduce SP4LP (Shortest Path for Link Prediction), a novel framework that combines GNN-based node encodings with sequence modeling over shortest paths. Specifically, SP4LP first applies a GNN to compute representations for all nodes, then extracts the shortest path between each candidate node pair and processes the resulting sequence of node embeddings using a sequence model. This design enables SP4LP to capture expressive multi-hop relational patterns with computational efficiency. Empirically, SP4LP achieves state-of-the-art performance across link prediction benchmarks. Theoretically, we prove that SP4LP is strictly more expressive than standard message-passing GNNs and several state-of-the-art structural features methods, establishing it as a general and principled approach for link prediction in graphs.

Problem

• GNNs fail to capture link-specific structural patterns for link prediction
• Existing methods are computationally costly or overly simplistic for multi-hop dependencies
• Propose SP4LP combining GNNs and sequence models for efficient multi-hop pattern capture

Innovation

• Combines GNNs with sequence models
• Uses shortest paths for link prediction
• Captures multi-hop relational patterns efficiently