Characterizing silent users in social media communities

Abstract

Silent users often constitute a significant proportion of an online user-generated content system. In the context of social media such as Twitter, users can opt to be silent all or most of the time. They are often called the invisible participants or lurkers. As lurkers contribute little to the online content, existing analysis often overlooks their presence and voices. However, we argue that understanding lurkers is important in many applications such as recommender systems, targeted advertising, and social sensing. This research therefore seeks to characterize lurkers in social media and propose methods to profile them. We examine 18 weeks of tweets generated by two Twitter communities consisting of more than 110K and 114K users respectively. We find that there are many lurkers in the two communities, and the proportion of lurkers in each community changes with time. We also show that by leveraging lurkers' neighbor content, we are able to profile them with accuracy comparable to that of profiling active users. It suggests that user generated content can be utilized for profiling lurkers and lurkers in Twitter are after all not that "invisible".