GED-Consistent Disentanglement of Aligned and Unaligned Substructures for Graph Similarity Learning

2025-11-24

AI Summary

Existing GNN-based GED approximation methods suffer from two key limitations: (i) difficulty in modeling global structural correspondences, and (ii) spurious signals introduced by node-level matching, leading to inaccurate edit cost estimation. This paper proposes a decoupled graph similarity learning framework—the first to jointly model graph-level alignment and substructure-level edit cost estimation. It explicitly distinguishes aligned from unaligned substructures, thereby avoiding structural mismatches and cost confounding. An end-to-end GNN architecture enables global alignment-aware similarity estimation. The method achieves state-of-the-art performance on four benchmark datasets. Ablation studies and visualization analyses confirm that the model learns semantically coherent, well-decoupled substructure representations—significantly improving both GED approximation accuracy and interpretability.

Technology Category

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

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Graph Similarity Computation (GSC) is a fundamental graph related task where Graph Edit Distance (GED) serves as a prevalent metric. GED is determined by an optimal alignment between a pair of graphs that partitions each into aligned (zero-cost) and unaligned (cost-incurring) substructures. Due to NP-hard nature of exact GED computation, GED approximations based on Graph Neural Network(GNN) have emerged. Existing GNN-based GED approaches typically learn node embeddings for each graph and then aggregate pairwise node similarities to estimate the final similarity. Despite their effectiveness, we identify a mismatch between this prevalent node-centric matching paradigm and the core principles of GED. This discrepancy leads to two critical limitations: (1) a failure to capture the global structural correspondence for optimal alignment, and (2) a misattribution of edit costs driven by spurious node level signals. To address these limitations, we propose GCGSim, a GED-consistent graph similarity learning framework centering on graph-level matching and substructure-level edit costs. Specifically, we make three core technical contributions. Extensive experiments on four benchmark datasets show that GCGSim achieves state-of-the-art performance. Our comprehensive analyses further validate that the framework effectively learns disentangled and semantically meaningful substructure representations.

Problem

• Addresses mismatch between node-centric matching and GED principles
• Resolves failure to capture global structural correspondence for alignment
• Corrects misattribution of edit costs from spurious node signals

Innovation

• Disentangles aligned and unaligned graph substructures
• Uses graph-level matching for similarity computation
• Models substructure-level edit costs for GED consistency