Discovery of topic based on mass incidents and research of user roles

Abstract

Based on the analysis of user behavior in social network, knowledge is introduced into social subject's behavior and explores the online social network public opinion research in the related method. First of all, based on the multi-dimensional social relationship between the users participating subject discussion, two-layer overlay network model of topic keywords with the user based on implicit link is put forward targeting at the problem of short text and semantic sparse in a new generation of online social network represented by Weibo system. The model takes the user as the center, the keyword as the basic unit, and applies the clustering method. The model can be used on the real data sets to show that all kinds of important users and user groups of the hot topics, and discuss different users and user group of the topics of concern in the form of a keyword to. It is found that the same role in different groups of users have different motives in the real social network. These findings have important practical significance for discovery and control of public opinion in the social network systems.