HITS Centrality Based on Inter-Layer Similarity for Multilayer Temporal Networks

Laishui Lv,Kun Zhang,Dalal Bardou,Xun Li,Ting Zhang,Wei Xue

DOI: https://doi.org/10.1016/j.neucom.2020.10.040

IF: 6

2021-01-01

Neurocomputing

Abstract:In this paper, we propose a novel HITS centrality based on inter-layer similarity for multilayer temporal networks, referred to as the MT-HITS centrality. We firstly introduce a sixth-order tensor to represent multilayer temporal networks and define an inter-layer similarity index to measure the strength of the interactions between different layers. Then, based on the established tensor, we construct a set of tensor equations to define centrality vectors of nodes, layers, and time stamps, respectively. These centrality vectors can be regarded as a Perron eigenvector of a multi-homogeneous map. Under some reasonable conditions, the existence and uniqueness of our proposed centrality measure are guaranteed by existing results. To demonstrate the effectiveness and superiority of our proposed method, we test it on three networks, and then we apply it for link prediction in temporal networks.