Influential nodes in complex networks no longer rely on one scale: local, community and global structure learned together, ranking beats centrality and GNN baselines
Synopsis
Influential-node identification in complex networks now has a cross-scale approach. Centrality-based and learning-based methods often integrate structural information across network scales only to a limited degree, so local, community-level and global patterns are not used together. CSBA-Net extracts structural representations at the three scales, models dependencies between them with a BiGRU and multi-head self-attention, and learns node influence scores by regression against SIR diffusion simulations; on 24 real and 12 synthetic networks its ranking metrics exceed the compared centrality, CNN and GNN baselines.
Interpretation
Influential-node ranking now draws on structural information at local, community-level and global network scales at the same time, rather than on a single scale of structural description. Existing centrality-based and learning-based methods often provide limited integration of structural information across different network scales. On 24 real-world and 12 synthetic networks from social, biological, technological and knowledge domains, the method reaches Kendall's tau=0.802, Spearman's rho=0.943, NDCG10=0.898 and NDCG20=0.916, outperforming the considered centrality-, CNN- and GNN-based baselines.
Cross-scale dependencies are captured by extracting structural representations at each scale separately and then modeling the relations between scales. The approach is the Cross-Scale Structural BiGRU Attention Network, CSBA-Net, a multi-scale structural representation learning framework. The extracted representations are processed by CNNs and integrated through a BiGRU with multi-head self-attention, while influential-node scores are learned by regression using SIR-based diffusion simulations as reference targets.
Perspective
The method targets influential-node identification in complex networks, evaluated on 24 real-world and 12 synthetic networks from social, biological, technological and knowledge domains, and it outputs node influence ranking scores. It suits researchers and engineers working on network analysis, diffusion modeling and graph learning.
Node influence scores are learned with SIR-based diffusion simulations as reference targets, so ranking performance is tied to the diffusion simulation setting used; the material does not state how it behaves under different spreading parameters.