A Tensor-Based Framework for Studying Eigenvector Multicentrality in Multilayer Networks

Authors:

Abstract
Multilayer networks have drawn much attention in the community of network science recently. Considerable effort has been invested to understand their structure and functions, among which centrality is identified as one of the most critical metrics. While various forms of centrality have been proposed for single-layer networks, a general framework for studying centrality in multilayer networks is lacking. Here we introduce a mathematical framework to study eigenvector multicentrality, which enables us to quantify the relationship between interlayer influence and eigenvector multicentrality, providing an analytical tool to describe how eigenvector multicentralities of nodes propagate among different layers. Further, the framework is flexible for integrating prior knowledge of the interplay among layers so as to attain a tailored eigenvector multicentrality for varying scenarios. We show how to model the multilayer influence by choosing appropriate influence weight functions and design algorithms to calculate eigenvector multicentrality in various typical scenarios. We apply this framework to analyze several empirical multilayer networks, finding that it can quantify the influence among layers and describe the structure-function relationship of multilayer networks very well.

This publication has 29 references indexed in Scilit: