CNBNews

Three researchers –  former Rutgers postdoctoral scholars Nick Diakopoulos and Munmun De Choudhury and Mor Naaman, an assistant professor in the School of Communication and Information – have created a tool they believe can help journalists find credible sources amid the din. 

“Traditionally, a journalist might have built up sources in a community. But now, online, who knows who that person tweeting information is?” said Diakopoulos, who led the project at Rutgers and is now an independent researcher and consultant. 

The system, known as Seriously Rapid Source Review, or SRSR (pronounced Sourcer), was developed and evaluated with reporters at seven news organization - Philly.com, the New York Daily News, NPR, The Huffington Post, The Washington Post, Reuters, and the Guardian. The work is funded by the National Science Foundation. 

The researchers began their project by talking to journalists about how they did their jobs. They examined how social media is used differently between hard and soft news, between breaking and continuing stories. They learned how journalists use, or would like to use, social media. They learned about the pressure of deadlines and what made a potential social media source credible in journalists’ eyes. Then they designed their prototype software system to meet reporters’ needs. 

SRSR presents journalists with a scrolling list of Twitter users who are potential news sources for a given event. Aside from their name, image, number of followers and other users followed, each entry includes other information about the users derived from other Twitter data – the user’s location, his or her relation to other users following the same event, connections to people in the geographic area, and qualifications. Is the user a journalist, a blogger, an expert of some sort? SRSR also sorts through the words in each user’s tweets to ascertain how likely he or she is to be an eyewitness to the event. 

Once the prototype system was developed, Naaman, Diakopoulos, and De Choudhury, who has since been hired by Microsoft Corp., installed the software on a laptop and asked journalists to give it a test drive, using Twitter data from two breaking news events of 2011, the riots in the Tottenham section of London, and the tornado that virtually destroyed Joplin, Missouri. They had data – 600 tweets about the riots, and 10,000 about the tornado. 

Generally, the editors found the system useful, though they had some suggestions for improvements. They liked SRSR’s information about a Twitter user’s connection to other people: the more connected the user, the more likely the journalists were to consider him or her a reliable source. Some wished that SRSR’s algorithm for gauging a source’s “eyewitness-ness” took the source’s location into account; others wanted the ability to build lists of likely sources based on their connectedness and locations. Throughout their discussions with the researchers, they emphasized that they were looking for “verity” – truth, in other words. 

“Our system doesn’t find the truth,” Naaman said. “It helps journalists find the truth by leading them to people who are in a good position to know the truth.” 

The researchers will present their work in May at the annual Conference on Human Factors in Computing Systems in Austin, Texas.