Extraction and Visualization of Implicit Social Relations on Social Networking Services

Abstract

Today social network services like blogs, communities, and social networking sites dominate the web. As Web 2.0 has evolved this way, analyzing social networks has become a promising research issue. There have already been several researches on social network analysis based on users' activities in social services. Most of them focus on the links among the users such as citation, trackback, and comment. However, few studies have analyzed the relations within message threads. In general, they considered the one-way relationship from a comment writer to a post writer. Since users communicate with each other primarily by posting comments one after another, the message threads are key to analyzing latent social relationships. In this paper, we propose a novel method to extract the social relations hidden behind message threads. To evaluate our algorithms, we developed an evaluation system and measured the performances. In addition, since the typical node-edge diagram for social network visualization is not intuitive or readable, we also introduce a novel visualization and interaction method suitable for social relation exploration. Further, we expect our work will help enhance social recommendations, advertisements and personalization.