Characterizing user influence within twitter

Abstract

This paper explores whether influence can be quantified from public Twitter data. Compared to other social media applications, Twitter is content-centered, rather than relationship-centered. There is no indication of mutual relationships for the user within the application, making it difficult to gauge influence. By analyzing the data that already had mutual relationships, we identify the characteristics that created the boundaries of a community, and influence within it. We looked at Twitter user data, as well as Tweet data to find ways to characterize user influence among them. We measure type of users based on factors such as: those that they follow and how active they are. The Expert members are mutually agreed upon, as evidenced by their large followings, and the large number of followers who have added them to a list. They are most likely to post replies and original tweets, and are unlikely to re-tweet. Active members keep the conversation going, as evidenced by their strong followings. They are more likely than the other types to re-tweet. Passive members, the largest group, participate by liking (Favorite) tweets that they consume, encouraging experts and active members to continue their actions, and sustaining the boundaries of the group.