实证网络中节点中心性的鲁棒性研究

Node centrality is an important quantity for complex networks as it is related to many applications ranging from the prediction of network structure to the control of dynamics on networks. In the literature, much effort has been devoted to design new centrality measurements. However, the reliability of these centrality measurements haven't been fully understood. Many real networks are facing different kinds of manipulations such as addition, removal or rewiring of links. In this paper, we focus on the robustness of classic centrality measures against network manipulation. Our analysis is based on both artificial and real networks. We find that the centrality measurements are generally more robust in heterogenous networks. Moreover, the top part of the centrality ranking is more resistant to manipulation. Among the centrality measures we considered, the eigenvector centrality is in general the most robust one.?????