Two MIT researchers have taught an algorithm to spot sarcasm by analyzing emoji usage.
(CCM) — An artificial intelligence system has been trained by two researchers to recognize sarcasm by looking and emoji usage, the BBC has reported. The researchers, from The Massachusetts Institute of Technology (MIT), trained DeepMoji by showing it 1.2 billion tweets which contained at least one of the 64 most commonly used emoji, according to the report. During the training DeepMoji learned to predict which emoji would likely be used in a tweet, and later it learned when the authors were being sarcastic. Many people, such as some people with autism, have difficulty detecting sarcasm, so the fact that a computer program can do so is impressive.
The ability to detect sarcasm is useful because it allows the artificial intelligence system to spot hate speech faster than humans, according to the report. This could allow social networks to police the content of their sites more efficiently. The technology could also be used to improve interactions between humans and automated bot software such as customer service chat bots.
Emoji are commonly used in internet communications, such as tweets and Snapchat messages, to contextualize what is being said in place of the voice intonations or body language used in verbal conversations.
The two researchers plan to open-source their artificial intelligence software so that it can be used by other researchers and interested parties in the future.
Image: © iStock.