Social Media Lie Detector: Can It Work ?

By Jolie Lee
A new project aims to figure out how to identify, in near real-time, whether they’re true.
Called Pheme, the three-year project is using text-mining technology to cull information from large amounts of text data, such as tweets, KalinaBontcheva, lead researcher for Pheme, said in an interview with USA TODAY Network.
The project is similar toan analysis conducted by The Guardian during the 2011 London riots, when the newspaper tracked how false rumors spread on Twitter, including one that animals were set free from the London Zoo.
The rumors will be classified based on several factors, including how they were spread and whether false statements were spread intentionally.

TwitterThe plan is to build Pheme as a Web application that would give users a confidence percentage of how likely a statement is true or false, said Bontcheva, a senior research scientist in computer science at the University of Sheffield.
Although false rumors can be rebutted online, sometimes wrong information still lives longer than it needs to on social media, according to Bontcheva.
The application could be used beyond a news context. For example, Pheme researchers are also testing how well the text-mining technology can figure out false statements when it comes to medical controversies.
The European Union is funding most of the project, with five universities and four companies developing it. Bontcheva said the researchers plan to have a prototype of Pheme in 18 months.
One of the project goals is to assist governments and agencies in times of emergency to disseminate accurate information, while debunking rumors that could panic the public, according to the developers.
But a digital rights advocate expressed concern.
“People should be skeptical,” said Dave Maass, spokesman for the Electronic Frontier Foundation, a not-for-profit digital rights group. “It isn’t the role of government to decide what is the truth.” He added that different governments have different interpretations of what the truth is.
Maass said the project also needs to make clear its privacy policy. Pheme will be looking at content on blogs, URLs contained in tweets, online forums and public Facebook pages.
“Non-public content is out of scope,” Bontcheva said in an e-mail.
Although the information is public, the amount of information collected could be “very invasive,” Maass said. He added that it will be important to know how long people using this technology will keep the information.
A social media expert pointed to the value of having information about the track records of users who are sharing information.
“Are there people who have strong, established reputations and have something to lose if it’s not true?” said Judith Donath, a fellow at the BerkmanCenter for Internet and Society at Harvard University. “These are useful pieces of information.”
But she added that there likely isn’t a perfect way of evaluating the truth of statements on social media.
“Tweets are hard for humans to understand because they’s so compact,” Donath said. “A computer isn’t necessarily going to do better.
Source: USA Today