A perennial problem in human-computer interaction is getting machines to understand emotional concepts like sarcasm, irony and understatement.
Now, however, researchers at the MIT Media Lab may have made a breakthrough. They've built a system that has scanned through 1.2 billion tweets with a natural emotional labelling system attached - emoji.
"Our basic idea with our DeepMoji project is that if the model is able to predict which emoji was included with a given sentence, then it has an understanding of the emotional content of that sentence," explains the team's project page.
"With this approach we beat the state of the art across benchmarks for sentiment, emotion, and sarcasm detection."
Try it yourself
You can try the software yourself by heading over to deepmoji.mit.edu and typing in a short sentence. The algorithm will predict which emoji should accompany it. It's pretty good at figuring out exactly what you mean.
"The classic use case is companies wanting to make sense of what their customers are saying about them," wrote the team.
"But there are many other use cases as well now that natural language processing (NLP) is becoming an increasingly important part of consumer products."
They continued: "For instance, all of the new chatbot services popping up might benefit from having a nuanced understanding of emotional content in text. Lastly, it can hopefully be used for various interesting research purposes."
To that end, the team plans to soon release the code so that it can be used by other researchers.
