Sarcasm detector invented for Twitter
Computers can now recognise when a social media user is being sarcastic, thanks to a new artificial intelligence system.
Twitter in particular is harvested by marketing firms, stock traders and intelligence services.
But because computer programmes cannot tell when a tweet is saying the opposite of what it actually means, sarcasm has thrown data systems.
That is now set to change, after Silvio Amir from the University of Lisbon created a machine learning technique to overcome the problem.
He began work in 2014 when the US Secret Service listed sarcasm detection as a much-needed feature in its social media analytics service.
His system works by combing through a user’s previous tweets to build up a picture of their character. It has an apparent accuracy level of 87 per cent, but that would improve over time.
Amir told the New Scientist: "It intuitively makes sense. Tell me what you talk about and I can tell you who you are.
"We can get away without looking at all this external information."
He added that the system could be adapted for any social media site where a user’s previous posts are visible.
In January this year, a programme called TrueRatr was developed which screens app reviews for sarcasm in order to create its own more accurate rating.
ArticleYeah right... Sarcasm detector invented for social mediasarcasmSilvio AmirTwitter