This could have been done a little more sophisticated. Now it seems nothing more than some basic text searches.
For instance two people saying "Not Hungover AT ALL, I love these mornings" and "Hungover !" respectively are both in the hungover section. And in the doing drugs section there was one guy who was happy he actually quit.
That's a nontrivial problem to solve. Consider "I avoided becoming hungover again!". In other words, you'd have to do a full semantic analysis and get the scope of negations right. I hear that current sentiment analysis techniques use a heuristic by looking for a nearby negation.
For instance two people saying "Not Hungover AT ALL, I love these mornings" and "Hungover !" respectively are both in the hungover section. And in the doing drugs section there was one guy who was happy he actually quit.