A RED-BET Method to Improve the Information Diffusion on Social Networks

Abstract

Information diffusion in the social network has been widely used in many fields today, from online marketing, e-government campaigns to predicting large social events. Some study focuses on how to discover a method to accelerate the parameter calculation for the information diffusion forecast in order to improve the efficiency of the information diffusion problem. The Betweenness Centrality is a significant indicator to identify the important people on social networks that should be aimed to maximize information diffusion. Thus, in this paper, we propose the RED-BET method to improve the information diffusion on social networks by a hybrid approach that allows to quickly determine the nodes having high Betweenness Centrality. Our main idea in the proposed method combines both the graph reduction and parallelization of the Betweenness Centrality calculation. Experimental results with the currently popular large datasets of SNAP and Animer have demonstrated that our proposed method improves the performance from 1.2 to 1.41 times compared to the TeexGraph toolkit, from 1.76 to 2.55 times than the NetworKit, and from 1.05 to 1.1 times in comparison with the bigGraph toolkit.