Talking Back: How Congress Engages With the Public on Twitter

This paper examines political social media interaction between Members of the United States Congress and the general public on Twitter. Specifically it attempts to gain a better understanding of what motivates a Member of Congress to reply to a tweet by building a model to predict which tweets Members of Congress will reply to. Predicting social media interaction of elected Members of Congress is a challenging machine learning task. MOCs reply more often to positive tweets based on personal interactions they have had and positive feedback. The distribution of reply to non-reply tweets makes predicting responses very error prone. In this process a several methods of predicting rare events in text were attempted with inconclusive results.Master of Science in Information Scienc