This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Find out more here

NEWS20 May 2010

Big whoop! Researchers develop online sarcasm detector

Data analytics North America

ISRAEL— Researchers at Hebrew University have developed a machine algorithim that can recognise sarcasm in online postings – but does it have the potential to add a new layer to social media sentiment analysis?

The algorithim was tested against 66,000 reviews for books and other products listed on Amazon, where it achieved 77% precision in identifying sarky comments.

Researchers trained the algorithim to learn the sorts of words and patterns that distinguish sarcastic comments before putting it to work.

“Sarcasm recognition could contribute to the performance of review summarisations and ranking systems,” write the researchers in their paper detailing the development of Sasi, or Semi-supervised Algorithim for Sarcasm Identification.

But what of social media analysis? Mark Westaby of Spectrum Consultancy, sees little commercial potential for such an algorithim.

Sarcasm can throw up false postives when analysing online sentiment, he says, but the market is increasingly demanding more than just basic assessments of whether a comment is positive or negative. “Insight”, he says, is what brand owners and marketers want.

Whether sarcastic or not, their aim is to get to the heart of what is being expressed – to shed light on what it is exactly that people like or dislike about a product or service.

“From an academic perspective, it’s very interesting that this work is being done,” says Westaby, “but as an application in the real world, I think it has very little going for it.”

@RESEARCH LIVE

2 Comments

9 years ago

Wow! This is a great idea! I look forward to seeing this put to action!

Like Report

9 years ago

“Insight”, he says, is what brand owners and marketers want." Exactly and that is why people need to move beyond mere monitoring and extact insight from what is being said, who is saying, where and with impact. From this insight you can extract actions to improve marketing ,communications, customer relations, product/service development, CSR, HR... And to do this, you need people in your webspace day in day out.

Like Report