MIT AI Project “DeepMoji” Can Detect Sarcasm on Social Media Via Emoji

We’ve been wanting a sarcasm font for the longest time, but it may not be needed, thanks to the brainy guys at MIT. They’ve developed an algorithm which can, supposedly, detect when someone’s being sarcastic on Twitter.

This is a result of a project that aims to detect and fight racism, abuse, and harassment using AI. As you can imagine, that project involves such a large scope that the researchers had to break down their goals into more digestible pieces. One of the things they realized along the way is that many messages on Twitter cannot be understood fully without grasping sarcasm.

As such, the MIT researchers developed the algorithm in such a way that it learns from data it studies. It’s already an existing machine-learning technique – expose a simulated neural network to large amounts of data and train it to recognize patterns.

In this case, sarcasm on social media is learned by the presence of emoji. The result? The algorithm called “DeepMoji”.

“Because we can’t use intonation in our voice or body language to contextualize what we are saying, emoji are the way we do it online,” says Iyad Rahwan, an associate professor the MIT Media lab who developed the algorithm with one of his students, Bjarke Felbo. “The neural network learned the connection between a certain kind of language and an emoji.” (Source)

As you may very well know, it can be a bit of hit and miss when it comes to recognizing sarcasm online. Some people are just not on the same wavelength, so it’s not improbable for DeepMoji to make mistakes. Still, when pitted against humans, DeepMoji won with an 82% accuracy rate (average for humans was 76%).