How rational are people? Economic behavior based on sentiment analysis

User-generated content plays a major role on the Web. It allows community websites to prosper as people seek each other´s opinions before making their own choice. In this paper, we propose a novel approach to investigate the correlation between the sentiments´ scores extracted from review text and the number of likes/dislikes for restaurants reviews, we also explore how the sentiments´ scores affect the rating of restaurants. We formalize our problem under the rational choice theory, in which a person would make a choice among set of alternatives that have different utility that is calculated based on some measures. We propose to use different combination of measures including the number of likes/dislikes, and different sentiments scores that are generated manually and automatically. Results show how top ten restaurants fluctuated when different utility measures were used. Based on our initial results, we encourage future research in applying economics´ theories on user-generated content to help in the process of sentiment analysis and decision making.