A Small Step into sarcasm detection

Detecting sarcasm is an active area of research (http://homes.cs.washington.edu/~nasmith/papers/bamman+smith.icwsm15.pdf). In fact, detecting sarcasm is often not easy for humans, so how can it be easy for computers? If I say "We will make America great again"; without knowing me, observing me, or hearing the tone I'm using, how could you know if I really meant what I said? Now, if you were to read a tweet from me that says "We will make America great again :(:(:(", does it help in a sense?

Building features

We believe that sarcasm cannot be detected using plain English text only, especially not when the plain text fits into less than 140 characters. However, we showed in this chapter that emojis can play a major role in the definition of emotion. A naive assumption is that a tweet with both positive sentiment and negative emojis can potentially lead to sarcasm. In addition to the tone, we also found that some words were closer to some ideas/ideologies that...