Diversity of online community activities

Abstract

Web sites where users create and rate content as well as form links display many long-tailed distributions. Using one such site, Essembly, we propose causal mechanisms to explain these behaviors. Unlike purely descriptive models, our mechanisms use only information available to each user. We find the long-tails arise from large diversity of user activity and qualities of the rated content. The models not only explain overall behavior but also allow estimating the qualities of users and content from their early history on the site.