Identifying multiple influential spreaders via local structural similarity

¹ Data Science and Cloud Service Centre, Shanghai University of Finance and Economics - Shanghai 200433, PRC
² Research Center of Complex Systems Science, University of Shanghai for Science and Technology Shanghai 200093, PRC
³ College of Humanities, Shanghai University of Finance and Economics - Shanghai 200433, PRC

^(a) liujg004@ustc.edu.cn

Received: 28 April 2017
Accepted: 9 August 2017

Abstract

Identifying the nodes with largest spreading influence is of significance for information diffusion, product exposure and contagious disease detection. In this letter, based on the local structural similarity, we present a method (LSS) to identify the multiple influential spreaders. Firstly, we choose the node with the largest degree as the first spreader. Then the new spreaders would be selected if they belong to the first- or second-order–neighbor node set of the spreaders and their local structural similarities with other spreaders are smaller than the threshold parameter r. Comparing with the susceptible-infected-recovered model, the experimental results for four empirical networks show that the spreading influences of spreaders selected by the local structural similarity method are larger than that of the color method, the degree, betweenness and closeness centralities. The further experimental results for the Barabàsi-Albert and random networks show that the LSS method could identify the multiple influential spreaders more accurately, which suggests that the local structural property plays a more important role than the distance for identifying multiple influential spreaders.