Bridge the Gaps: Heterogeneous Attributed Graph Clustering via Quaternion Representation Learning

2026-06-22

AI Summary

This work addresses the challenges of clustering heterogeneous attributed graphs, where numerical and categorical attributes coexist and graph learning often suffers from representation degeneration—such as oversmoothing and excessive topological dominance. To tackle these issues, the authors propose AGREE, an end-to-end framework that introduces quaternion graph convolution to enhance interactions among diverse attribute types for the first time. AGREE unifies multi-level alignment and similarity-driven graph construction within a shallow architecture to mitigate oversmoothing, while jointly optimizing graph reconstruction and clustering objectives. Notably, it operates without requiring a predefined number of clusters. Extensive experiments demonstrate that AGREE significantly outperforms state-of-the-art methods in terms of clustering accuracy, robustness, and adaptability across multiple benchmark datasets.

Abstract

Attributed graph clustering partitions nodes by jointly exploiting node attributes and graph topology. It remains challenging due to attribute heterogeneity and representation degradation during graph learning. Real-world datasets often contain heterogeneous attributes, i.e., numerical and categorical attributes, complicating unified representation learning. This challenge becomes more complex in attributed graphs, where constructing a clustering-friendly graph structure from attributes and topology remains difficult. Under deep graph architectures, repeated graph propagation causes node embeddings to become overly similar, leading to the over-smoothing (OS) effect. Meanwhile, graph representation learning amplifies topological influence, making discriminative attribute information harder to exploit for clustering, an effect we refer to as over-dominating (OD). To bridge these gaps, an end-to-end framework, Any-type attributed Graph REpresentation lEarning (AGREE), is proposed. It unifies attributed graphs and any-type attributed data through multi-level alignment and similarity-based graph construction. Quaternion-based graph convolution strengthens attribute interaction to alleviate OD, while shallow graph architectures help relieve OS. The learned embeddings are jointly optimized for graph reconstruction and clustering, without requiring a predefined number of clusters during training. Experiments on diverse benchmarks show that AGREE achieves strong overall performance in accuracy, robustness, and adaptability.

Problem

• attributed graph clustering
• attribute heterogeneity
• over-smoothing
• over-dominating
• representation learning

Innovation

• Quaternion Representation
• Heterogeneous Attributed Graph
• Over-smoothing Mitigation
• Over-dominating Alleviation
• End-to-end Clustering