A multi-granularity collective behavior analysis approach for online social networks

Abstract

Online social networks play more and more important roles in the modern society in terms of the rapid and large scale information spread. Many efforts have been made to understand these phenomena in the computer science communities and other relative fields, ranging from popular topic detection to information diffusion modeling. In this article, a multi-granularity collective behavior modeling approach has been proposed, which describes the collective behavior at a node level, a neighbor level, a community level, and a social level. How collective behaviors on the largest online social network Weibo are formed is analyzed by this model. The results demonstrate that individuals can rationally make their decisions when they are isolated from others. However, when neighbors exist, their actions will be irrational; the phenomena will be more obvious when social reinforcement is taken into consideration. Members in the same community can easily reach a group consensus while holding different opinions in different communities.