A new centrality measure for probabilistic diffusion in network

Due to the significant increment of the volume of interactionsamong the population, probabilistic process on complex networkcan be often utilized to analyse diffusion phenomena in thesociety, then a number of researchers have studied especiallyfrom the perspectives of social network analysis, computer virusspread study, and epidemics study. So far, it has been believedthat the largest eigenvalue and the principal eigenvector of theadjacency matrix can well approximate the dynamics onnetworks, but the accuracy of this approximation method has notstudy extensively. In our previous work, we found that not onlythe largest eigenvalue and the principle eigenvector but also theother eigenvalues and eigenvectors need to be considered whenanalysing the diffusion process on real networks. In this paper,we proposed a new centrality measure, the infection diffusioneigenvector centrality (IDEC), which considers all eigenvaluesand eigenvectors. Our comparison results indicates that the IDECshows better predictability than other centrality measures whenthe effective infection ratio is low, which will provide us with agood insight for practical application for developing the effectiveinfection prevention methodology. Also, another interestingfinding is that the eigenvector centrality shows poorpredictability especially on the real networks. In addition, weconduct the recovery probability enforcement simulation, whichhighlights the advantage of IDEC for the range below the criticalpoint

A new centrality measure for probabilistic diffusion in network Keywords:

Infection , SIS model , Complex network , Centrality , Eigenvalue , Eigenvector