A classifier that distinguishes between romantic and nonromantic posts using topic model derived features, romance lexicons inspired by sentiment analysis, and features extracted about the subject(s) of the post and the post itself.
Results: The best decision tree classifier run achieved classification accuracy of 83% on the test set. Our best logistic regression classifier run achieved a classification accuracy of 70% on the test set.
Collaboration with Ian Gonzalez & David McPeek at Yale.
