A new algorithm developed by professor Iyad Rahwan and graduate student Bjarke Felbo called ‘Deepmoji’ has employed the use of Emojis to analyse the use of sarcasm online. The team used 1.2 billion tweets containing one of the 64 most widely-used emojis in their research.

Through their research, the algorithm has learnt how to predict the most likely emoji to be used, and then furthermore, and rather more impressively, whether the tweet is genuine, or if it has sarcasm laced into it. Although this may seem purely humorous at first, this does have security implications, as it will allow the program to spot hate speech much faster than a human would.

Kerstin Dautenhahn from the University of Hertfordshire thinks that “Using emojis as labels for training neural networks is a great idea… Applying it to tweets seems also a smart choice, since communication via tweets is much more impoverished than actual face-to-face conversation”

The United Kingdom is famed for its sarcastic sense of humour, and therefore it must be a shock to know that something as technical and non-personal as a computer algorithm can correctly predict sarcasm by a small face expressing basic emotions.

Emojis were developed in Japan in the late 1990’s; the last decade has seen these little yellow faces spawn into one of the most commonly used phenomena in the digital age. The simple smiley faces have turned into a whole host of expressive pictures, from an astronaut to an octopus, each with both obvious and sometimes underlying meanings (we’re looking at you, aubergine!)

Modern language students may want to look at the way that communication is changing, and the different languages that come about from the digital age. Computer Science students may want to look at the way in which algorithms are being developed to suit popular culture. Psychology students may want to think about how complex human emotions can be condensed into a few brightly coloured pixels.

More Resources