CLGNN: A Contrastive Learning-based GNN Model for Betweenness Centrality Prediction on Temporal Graphs

2025-06-17

AI Summary

Addressing the high computational cost of exact temporal betweenness centrality (TBC) computation and the severe label imbalance (predominantly zero values) that hinders model identification of critical nodes in temporal networks, this paper proposes KContrastNet—the first TBC prediction framework integrating stability-driven clustering-guided contrastive learning with a dual-path aggregation graph neural network (GNN). Methodologically, it introduces instance-graph modeling, dual-path aggregation via mean pooling and edge-to-node multi-head attention, time encoding, path-counting enhancement, and support for multi-semantic optimal path definitions. Experiments demonstrate that KContrastNet achieves up to 663.7× faster inference than exact algorithms, reduces mean absolute error (MAE) by 31.4× compared to static GNNs, and improves Spearman correlation by 16.7×, significantly outperforming state-of-the-art temporal GNN approaches.

Technology Category

• Planning, Routing, and Scheduling: Temporal Planning
• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

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

Temporal Betweenness Centrality (TBC) measures how often a node appears on optimal temporal paths, reflecting its importance in temporal networks. However, exact computation is highly expensive, and real-world TBC distributions are extremely imbalanced. The severe imbalance leads learning-based models to overfit to zero-centrality nodes, resulting in inaccurate TBC predictions and failure to identify truly central nodes. Existing graph neural network (GNN) methods either fail to handle such imbalance or ignore temporal dependencies altogether. To address these issues, we propose a scalable and inductive contrastive learning-based GNN (CLGNN) for accurate and efficient TBC prediction. CLGNN builds an instance graph to preserve path validity and temporal order, then encodes structural and temporal features using dual aggregation, i.e., mean and edge-to-node multi-head attention mechanisms, enhanced by temporal path count and time encodings. A stability-based clustering-guided contrastive module (KContrastNet) is introduced to separate high-, median-, and low-centrality nodes in representation space, mitigating class imbalance, while a regression module (ValueNet) estimates TBC values. CLGNN also supports multiple optimal path definitions to accommodate diverse temporal semantics. Extensive experiments demonstrate the effectiveness and efficiency of CLGNN across diverse benchmarks. CLGNN achieves up to a 663.7~$ imes$ speedup compared to state-of-the-art exact TBC computation methods. It outperforms leading static GNN baselines with up to 31.4~$ imes$ lower MAE and 16.7~$ imes$ higher Spearman correlation, and surpasses state-of-the-art temporal GNNs with up to 5.7~$ imes$ lower MAE and 3.9~$ imes$ higher Spearman correlation.

Problem

• Predicts Temporal Betweenness Centrality efficiently on imbalanced graphs
• Addresses class imbalance in learning-based TBC prediction models
• Handles temporal dependencies ignored by existing GNN methods

Innovation

• Contrastive learning GNN for TBC prediction
• Dual aggregation with temporal path encoding
• Stability-based clustering mitigates class imbalance