Hunting for influential nodes based on radiation theory in complex networks

Assessing node importance is a crucial and long-standing research topic in the field of complex systems, which is vital for studying the robustness of networks and controlling network propagation. In our study, we propose a novel approach to identify influential nodes in networks by applying interdisciplinary knowledge from attenuation and scattering theories in atmospheric radiation physics. We use the attenuation theory to model information dissipation on the propagation path and the scattering theory to define information loss or gain at network nodes. Our new indicator for node importance assessment, called radiation centrality, can define the influence of network structure on node importance globally or globally-locally, depending on different parameterization schemes. To validate our approach, we rank the importance of nodes in different types of simulated networks and real networks with varying sizes using eight classical vital node identification algorithms and radiation centrality jointly. We compare these results with the node importance ranking results obtained from network propagation experiments, and apply the ranking results to network disintegration simulation experiments. We also validate our approach on a cooperative network of scientists with self-importance properties crawled by the network. Experimental results disclose the effectiveness of the radiation centrality algorithm and demonstrate that our approach can effectively distinguish the differences between nodes.