Interaction-Based Social Relationship Type Identification in Microblog

Qiao Deng, Zhoujun Li, Xiaoming Zhang, Jiali Xia

Abstract: Relationships in Microblogging services are lack of explicit meaningful labels, such as “colleagues”, “family members”, etc. The state-of-the-arts mainly work on mining only one particular relationship type such as advisor-advisee relationship for specific social networks. Moreover, few work focuses on relationship identification in Microblog based on link analysis. In Microblog, words in interactive tweets between users may provide clues for relationship type identification. In this study, we propose a two-step framework to infer the different social relationship types between users in Microblog. Firstly, a generative model UIRCT (User Interaction-based Relationship-related Community Topic) is proposed to discover relationship-related communities based on interactive content between users. We then profile the discovered communities with different relationship type labels by utilizing external resource. Experiment results on Sina Weibo dataset demonstrate that our proposed framework can identify different meaningful relationship types effectively.