Inter-layer similarity-based eigenvector centrality measures for temporal networks

Identifying the influential nodes in temporal networks has attracted lots of attention recently. In this paper, we present an Improved Eigenvector-based Centrality Measures (IECM) for temporal networks by regarding the coupling strength between proximity layers as the inter-layer similarity. Compared with the results of the nodes' influences got by temporal global efficiency for two real networks, the IECM method could identify influential nodes more accurately than the traditional ECM method. Regarding to the fact that different kinds of measurements have different performances, we introduce the Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) method to measure the global performance. Specially, when the inter-layer coupling strength ω set as 1 in the ECM method, the accuracy could be averagely enhanced 18.75% and 29.65% at each time layer for Workspace and Enrons datasets respectively, which indicates that measuring the inter-layer coupling strength plays an important role for identifying the influential nodes.