People often have a hard time understanding sarcasm, leading to the frequent breakout of inadvertent flame wars on the internet. Now imagine how difficult this is for computers, which tend to perceive things in the most literal way possible. However, researchers at MIT have developed a sentiment analysis technology called DeepMoji that can detect sarcasm better than humans can by using emojis.
The MIT team originally set out to develop technology that could identify racist statements on Twitter, according to the MIT Technology Review. However, the researchers quickly became aware that the system needed to be able to understand sarcasm before it could make a judgment the true sentiment of a statement.
To accomplish this, the MIT researchers utilized deep-learning techniques that employed a neural network for sentiment analysis. For training, the network was exposed to more than 1 billion posts on Twitter. Each of the tweets included one of 64 commonly used emojis, including smiling faces, frowning faces, hearts and a hand displaying a peace sign.
The emojis served as a sort of emotional guidepost for the neural network, helping it to understand the sentiment underlying the words in the tweet.
“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,” said Iyad Rahwan, an associate professor the MIT Media lab who co-developed the algorithm. “The neural network learned the connection between a certain kind of language and an emoji.”
When benchmarked using sentiment analysis tools, DeepMoji outperformed existing algorithms, according to the MIT team. More significantly, tests proved DeepMoji is better at identifying sarcasm and other emotions better than people can. The neural network was 82 percent accurate in detecting sarcastic statements, compared to just 76 percent for a group of human volunteers, according to the MIT research team.
While emojis are sometimes derided as evidence of a decline of literacy and writing skills, the MIT news illustrates the increasingly important role the symbols are play in online communications. Some experts even believe that emojis represent a new language that’s resurrecting ancient pictographic forms of writing, like Egyptian hieroglyphics.
“As a visual language emoji has already far eclipsed hieroglyphics, its ancient Egyptian precursor which took centuries to develop,” said professor Vyv Evans of Bangor University in comments reported by the BBC. “Emoji is the fastest growing form of language in history based on its incredible adoption rate and speed of evolution,” he added.
Nirel Marofsky is project analyst for the cognitive engine and application ecosystem at Veritone. She acts as a liaison to strategic partners, integrating developers and their capabilities into the Veritone Platform. Learn more about our platform and join the Veritone developer ecosystem today.