Predicting literature’s early impact with sentiment analysis in Twitter