Sarcasm meter 2.0

French company Spotter has developed an analytics tool that claims to have up to 80% accuracy in identifying sarcastic comments posted online.

Spotter says its clients include the Home Office, EU Commission and Dubai Courts.

The algorithm-based analytics software generates reputation reports based on social and traditional media material.

However some experts say such tools are often inadequate because of the nuance of language.

A spokeswoman for the Home Office said she should not comment at this time.

Spotter's UK sales director Richard May said the company monitored material that was "publicly available".

Its proprietary software uses a combination of linguistics, semantics and heuristics to create algorithms that generate reports about online reputation.

The company says these reports can also be verified by human analysts if the client wishes.

Algorithms had been developed to reflect various tones in 29 different languages including Chinese, Russian and Arabic, said Mr May.

"Nothing is fool-proof - we are talking about automated systems," he told the BBC.

"But five years ago you couldn't get this level of accuracy - we were at the 50% mark."

Mr May added one of the most common subjects for sarcasm was bad service - such as delayed journeys.

"One of our clients is Air France. If someone has a delayed flight, they will tweet, 'Thanks Air France for getting us into London two hours late' - obviously they are not actually thanking them," he said.

"We also have to be very specific to specific industries. The word 'virus' is usually negative. But if you're talking about virus in the context of the medical industry, it might not be."

Spotter charged a minimum of £1,000 per month for its services, Mr May said.

Human effort
Simon Collister, who lectures in PR and social media at the London College of Communication, told the BBC there was "no magic bullet" when it came to analytics that recognise tone.

"These tools are often next to useless - in terms of understanding tone, sarcasm, it's so dependent on context and human languages," he said.

"It's social media and what makes it interesting and fascinating is the social side - machines just can't comprehend that side of things in my opinion."

Mr Collister added that human interpretation was still vital.

"The challenge that governments and businesses have is whether to rely on automated tools that are not that effective or to engage a huge amount of human effort."


http://www.bbc.co.uk/news/technology-23160583