Towards Tracking Political Sentiment through Microblog Data

People express and amplify political opinions in Microblogs such as Twitter, especially when major political decisions are made. Twitter provides a useful vehicle for capturing and tracking popular opinion on burning issues of the day. In this paper, we focus on tracking the changes in political sentiment related to the U.S. Supreme Court (SCOTUS) and its decisions, focusing on the key dimensions on support, emotional intensity, and polarity. Measuring changes in these sentiment dimensions could be useful for social and political scientists, policy makers, and the public. This preliminary work adapts existing sentiment analysis techniques to these new dimensions and the specifics of the corpus (Twitter). We illustrate the promise of our work with an important case study of tracking sentiment change building up to, and immediately following one recent landmark Supreme Court decision. This example illustrates how our work could help answer fundamental research questions in political science about the nature of Supreme Court power and its capacity to influence public discourse. 1 Background and Motivation Political opinions are a popular topic in Microblogs. On June 26th, 2013, when the U.S. Supreme Court announced the decision on the unconstitutionality of the ”Defense of Marriage Act” (DOMA), there were millions of Tweets about the users’ opinions of the decision. In their Tweets, people not only voice their opinions about the issues at stake, expressing different dimensions of sentiment, such as support or opposition to the decision, or anger or happiness. Thus, simply applying traditional sentiment analysis scales such as ”positive” vs. ”negative” classification would not be sufficient to understand the public reaction to political decisions. Research on mass opinion and the Supreme Court is valuable as it could shed light on the fundamental and related normative concerns about the role of constitutional review in American governance, which emerge in a political system possessing democratic institutions at cross-purposes. One line of thought, beginning with Dahl (Dahl, 1957), suggests that the Supreme Court of the United States has a unique capacity among major institutions of American government to leverage its legitimacy in order to change mass opinion regarding salient policies. If the Dahl’s hypothesis is correct, then the Supreme Court’s same-sex marriage decisions should have resulted in a measurable change in opinion. A primary finding about implication of Dahl’s hypothesis is that the Court is polarizing, creating more supportive opinions of the policies it reviews among those who supported the policy before the decision and more negative opinions among those who opposed the policy prior to the decision (Franklin and Kosaki, 1989) (Johnson and Martin, 1998). We consider Twitter as important example of social expression of opinion. Recent studies of content on Twitter have revealed that 85% of Twitter content is related to spreading and commenting on headline news (Kwak et al., 2010); when users talk about commercial brands in their Tweets, about 20% of them have personal sentiment involved (Jansen et al., 2009). These statistical evidences imply that Twitter has became a portal for public to express opinions. In the context of politics, Twitter content, together with Twitter users’