Institute of Sports Science

Natural language processing for identifying fans from Twitter data

Previous research exploring the role of belief dynamics for consumers in the entertainment industry has largely ignored the fact that emotional reactions are a function of the conntent and a consumer’s disposition towards certain protagonists. By analyzing 19m tweets in combination with in-play information for 380 football matches played in the English Premier League we contribute to the literature in three ways. First, we present a setting for testing how belief dynamics drive behavior which is characterized by several desirable features for empirical research. Second, we present an approach for detecting fans and haters of a club as well as neutrals via sentiment revealed in Tweets. Third, by looking at behavioral responses to the temporal resolution of uncertainty during a game, we offer a fine-grained empirical test for the popular uncertainty-of-outcome hypothesis in sports.

Research Line: Sports Consumer Behavior.

Funding / Support:  #DataGrantsscheme by Twitter, Research Grant by the School of Business, Economics and Informatics (Birbeck University of London).

Principal Investigators: Dooruj Rambaccussing (University of Dundee), James Reade (University of Reading), Giambattista Rossi (Birkbeck University of London), Tim Pawlowski (University of Tübingen)

Publication: