Inferring FOLLOW Relationship from Repost Relationship between Users on Sina Weibo

Abstract

In recent years, online social platforms such as sina weibo have been playing an increasingly important role in our lives. With the improvement of public awareness of privacy protection, however, it is increasingly difficult to obtain exhaustive direct data of FOLLOW relationship between users, which restricts the analysis and research on real online social networks. In the context of incomplete disclosure of information, we have to consider inferring from limited public data to more unknown information before further analysis and research with network topology. To overcome this obstacle, we try to recover the FOLLOW relationship by repost relationship. We collect repost data of six different bloggers on sina weibo to generate six independent networks, and propose an effective method to recover FOLLOW relationship between users. Our method is superior to other existing methods, and helps us draw a conclusion that two users establishing an indirect repost relationship with other two users, compared to other indirect relationship, are of greater possibility to maintain the FOLLOW relationship. At the same time, we also extend our method to greatly improve the recovering accuracy.