Examining Multi-factor Interactions in Microblogging Based on Log-linear Modeling

Microblogging, as a new form of social media, attracts a huge number of users and becomes very popular. In this paper, we consider a fundamental social network issue that illustrates how information flows through a social media network and specify why users have different retweet behaviors. We propose to characterize social ties by using various features such as power ratio, local link structure, location, and gender. Those features can be directly extracted from users' profiles in Microblogging sites. We apply a fitted Log-linear model to describe association patterns among the features and retweet factor. Using the fitted Log-linear model, we explain why users with different profiles and link structures have different retweet behaviors. Our evaluations on Sina Weibo data set show several phenomenons.