Identify Influential Spreaders in Complex Real-World Networks

Identifying the most influential spreaders in a complex network is important in optimizing the use of available resource and controlling spreading behaviors on it. Centrality is usually used to measure the importance of a node within the network, such as degree, betweenness, closeness, eigenvector, k-core, etc. Here considering the local connection pattern of nodes in the network structure, we propose a new centrality measure which is based not only on the nearest neighborhood of a node, but also on its 2-step and 3-step neighbors. To evaluate its effectiveness, we use the classic spreading model to simulate the spreading efficiency of nodes in the network and compare the performance of the proposed centrality with the most widely used centrality of degree and coreness in ranking spreaders. Results show that the proposed centrality is a much more accurate measure to predict spreading capability of nodes in real-world networks.