Influential nodes identification for complex networks based on multi-feature fusion

Dentifying critical nodes in complex networks presents a significant challenge that has garnered extensive research attention. Previous studies often overlook the importance of spatial information, thereby limiting the accurate identification of key nodes. To address this gap, we introduce an advanced centrality model, termed Degree-k-shell-Betweenness Centrality (DKBC), which is grounded in the principle of gravity. The DKBC model integrates the centrality attributes of node degree, spatial positioning, and intermediate degree, resulting in improved accuracy for key node identification in complex networks. This innovative approach outperforms traditional gravity-based methods in terms of effectiveness. We validated the diffusion capacity of the proposed model using the Susceptible-Infected-Recovered (SIR) epidemic model and the Independent Cascade (IC) model, assessing correlation through the Kendall coefficient \(\tau\). A comparative analysis with benchmark algorithms highlights the superior performance of the DKBC model. Empirical validation across twelve real-world networks demonstrates the model’s exceptional accuracy in identifying key nodes. This study significantly advances the field by illustrating the effectiveness of incorporating spatial information into centrality measures to enhance both network analysis and practical applications.