Understanding prominence on massive and anonymous social news sites

Abstract

Many algorithms have been proposed to measure prominence of individuals in social networks. However, these algorithms are highly dependent on the underlying mechanisms for judging how individuals gain prominence in a given network. While there is a great deal of study in measuring prominence in networks of collaborations such as in academic networks and movie industry, there is relatively little work in understanding prominence in new and emerging social media sites. In particular, while there is some work in understanding the importance of friends and followers in Twitter, vote behavior in sites like Epinions, Digg and Slashdot, there is almost no study to this date on sites that encourage anonymous interactions like Reddit.

In this thesis, we explore the properties of interactions on Reddit. Reddit is different in that content is given either an "upvote", "downvote", or no vote from each user. The sum of the upvotes a user gets minus the downvotes is that user's Karma. Many users believe that Karma is an indicator of their prominence on Reddit, and others believe it is too noisy to be an effective measure. We evaluate how well a current algorithm based on collaborations, and several graph-based measures perform in finding prominent users, and how these results compare to various functions of Karma. We discuss what the performance of these algorithms reveal about the processes that lead to prominence on Reddit and how well Karma serves as a measure prominence.