Twitter is great for disseminating information quickly to millions. It's also great for disseminating misinformation. Now a team of European researchers is working to make the Twittersphere more reliable — by identifying lies before they go viral.
This seemingly impossible task will be accomplished by software called Pheme, a program intended to identify Twitter rumors in times of crisis.
See also: 10 Must-Follow Breaking News Accounts on Twitter
Pheme will be able to identify false information by looking at the news source, the conversations that stem from the tweet, and even the tweet's language, says senior researcher Kalina Bontcheva, an expert in text mining at Sheffield University. For example, researchers hope, Pheme will be able to identify the kind of sensationalist language or strong emotion commonly employed when people exaggerate.
The researchers, which include five universities and four different companies, are also looking at historical data to determine which rumors have spread in the past, and which users might be spam bots (accounts operated by machines, not actual people).
The project was inspired by University of Warwick professor of social informatics Rob Procter, who examined the spread of rumors on Twitter following the London riots in 2011. (Warwick is another university participating in Pheme.)
Procter's study covered claims that rioters released animals from the London Zoo and started to burn the London Eye, a giant ferris wheel on the River Thames. Neither rumor was true, but both were perpetuated on Twitter before other users discredited the claims.
Procter found that Twitter eventually corrects itself — there are enough people on the service who work to squash a rumor, but it's not necessarily quick. Bontcheva believes Pheme could help silence the rumors faster, and serve as a tool for journalists as they try to sort fact from fiction in a breaking news situation.
The idea is that Pheme would send information to a dashboard that journalists would monitor to determine the legitimacy of news on Twitter. "[The journalists] will still have to go the extra mile in the end," she says. "We are trying to help them get an overview of what is happening ... how is the conversation going?"
Bontcheva says Pheme will be most useful in times of crisis where many citizen journalists are contributing news, but that Pheme could also help in other situations — such as verifying medical information and helping to track the spread of diseases.

The London Eye. A photo-shopped picture of this ferris wheel created a Twitter frenzy during the London Riots in 2011.
Image: Flickr, Urko Dorronsoro
"We're not trying to police the Internet," she says. "We're not going to create a platform which looks through Twitter all the time and then say 'this tweet needs to be deleted or this person's account needs to be taken away.' It's not going to be like that because obviously I don't think anybody has the authority to do that."
The project isn't perfect, of course. For starters, Pheme doesn't recognize images or video, and a photoshopped image was what originally sparked the London Eye rumors. The tools is also relatively far off — it will be 18 to 20 more moths until the first prototype, says Bontcheva.
Still, she believes that the end result will make Twitter a more accurate place for news. And in times of crisis, that could make all the difference.
Have something to add to this story? Share it in the comments.