Tweeting Apart: Applying Network Analysis to Detect Selective Exposure Clusters in Twitter

Abstract

Twitter users see content mostly from the other users they select to follow. Networks of connected users on Twitter define the set of content to which each user is exposed. We developed a Selective Exposure Cluster (SEC) method to study these connected networks and their discussion patterns in Twitter. To illustrate the SEC method, we collected networks of connections among users who talked about a shared topic: the U.S. President's State of the Union speech in 2012. Cluster analysis was applied to identify subgroups of users who were densely interconnected. These users followed users from their own cluster more than they connected to users in other clusters. In each cluster, the primary distributors of information—the hub users—were identified, along with the hashtags, hyperlinks, and top-mentioned usernames in the content of the messages. Each of these indicators was labeled in terms of its political orientation. An analysis of the resulting patterns of selective exposure suggests that users participate in fragmented interactions and form divided groups in which people tune in to a narrow segment of the wider range of politically oriented information sources. We discuss the strengths and weaknesses of the proposed Selective Exposure Cluster method.