Using Clustering to Explore Teacher Communication on Twitter

While public school teachers in the United States may be constrained from expressing opinions about topics relating to their jobs in some public venues, Twitter provides a social platform where these teachers can discuss teaching-related issues and moral views as private citizens. In this work, we develop a computational system, Tweet Capture and Clustering System (TCCS), to support exploration of teachers using Twitter. In particular, TCCS looks at the tweets of a group of teacher tweeters and a list of words of interest, and seeks to identify subgroups of these teacher tweeters who have similar word-usage patterns. In the study reported here, we gather tweets from public Twitter accounts of self-identified teachers over an 11-month period. Through the use of TCCS, we have identified five distinct clusters of teacher tweeters, based on their word usage. Three of these five clusters are defined by the use of moral words. We find that the development and application of computational tools and methods, namely TCCS, allows the exploration of complex philosophical topics in public communication about education.