Preferential behavior in online groups

摘要: Online communities in the form of message boards, listservs, and newsgroups continue to represent a considerable amount social activity on Internet. Every year thousands groups ourish while others decline into relative obscurity; likewise, millions members join new community every year, some whom will come manage or moderate conversation simply sit by sidelines observe. These processes group formation, growth, dissolution are central science, an online venue they have ramifications for design development softwareIn this paper we explore large corpus thriving communities. vary widely size, moderation privacy, cover equally diverse set subject matter. We present broad range descriptive statistics these groups. Using metadata from groups, members, individual messages, identify users who post replied-to frequently multiple members; classify high-engagement based longevity their engagements. show that go become long-lived, highly-engaged experience significantly better treatment than other moment group, well before there is opportunity them develop long-standing relationship with groupWe simple model explaining long-term heavy engagement as combination user-dependent group-dependent factors. analytical tool, properties user alone sufficient explain 95% all memberships, but introducing small per-group information dramatically improves our ability belonging