Topical Engagement on Twitter: Using Consistency of Activity as a Means of User Segmentation
Keywords:Twitter, Politics, User behavior
In this paper we introduce a measure for the consistency of Twitter user behavior as a way of measuring engagement with a topic. Using this measure of topical engagement, we filter tweets collected from over 90,000 Nigerian users over a three-year period concerning that country’s head of state, President Goodluck Jonathan. We show that this measure effectively identifies a small set of highly engaged users that produce a disproportionate amount of activity, both in tweet volume and mentions of other users. Additionally, we show preferential within-group activity, with these users disproportionately mentioning other highly engaged users. Lastly, we show the potential of using changes in the percentage of engaged users within the overall user set for gauging the significance of events that occurred in Nigeria over the 2011 to early 2013 period.