Identifying influential spreaders in complex networks using neighbourhood coreness and path diversity

Abstract

The k-shell decomposition method dividing a great deal of nodes with different propagation capabilities into the same k-shell layer is unable to identify the influential spreaders accurately. Previous works improving the k-shell centrality were promising but inadequate, due to local neighbourhood and spreading dynamics of information. To solve this problem, the path diversity based on information entropy is proposed. We have investigated the spreading dynamics using susceptible-infected model and independent cascade model to reveal the behaviour of influential spreaders on the basis of topological location and neighbourhood information. Accordingly, a novel neighbourhood coreness method using path diversity to identify the influential spreaders from the point of information dissemination is proposed in this work. The simulation is evaluated with two real network datasets. The experimental results show that the neighbourhood coreness centrality with the spreading diversity is capable of identifying the influential spreaders more effectively and rank the spreading influence in a more fine-grained level. The nodes found by our method can produce a wider spreading scope in independent cascade model and can take less time to achieve the saturation point in susceptible-infected model.