Analysing the Structure, Dynamics and Contents of Social Networks

Abstract

Social media permits its users to have numerous accounts in diverse social networking podiums. Popular social networking sites have billions of users that possess similar interests. The aim of this paper is to determine the identity of users and map those identities with those having similar interests. The developed system can be utilized for business intelligence that would help in creating a database of all users that possess specific interests. In this study, a system is developed wherein a user's core interest is recognized in well-known social networking sites (i.e., Twitter and Plurk) by means of latent Dirichlet allocation (LDA) algorithm and Kullback-Leibler divergence method. In addition, in order to identify users across the abovementioned social networking sites, LDA will cogitate social network structure and article content of users. LDA initially establishes a user's core interest. Next, it calculates similarity between target users. The results in this study prove to be quite promising, which reveal that LDA is an effective algorithm to map users across various platforms and efficient when compared with other methods.