Twitter is one of the biggest social networking tools to get a message across.
However, according to researchers at Georgia Institute of Technology, specific words or phrases can identify a level of perceived credibility on Twitter.
While analyzing 66 million tweets, associated with 1,400 events, researchers were able to determine that comments from many social media users can impact a major event’s credibility, even if it’s ongoing, among other factors, PsyPost reports.
According to Tanushree Mitra, the lead researcher, over the past several years, social media credibility has been widely speculative, despite numerous studies.
However, the Georgia Tech Ph.D. candidate aimed at understanding which type of words or phrases adds the perception of credibility to suddenly transpiring events.
Between 2014 and 2015, researchers scattered through tweets of major events covered by the media such as the rise of Ebola, Charlie Hebdo attack and the backlash associated with the death of Eric Garner.
The credibility of the posts was then judged from “certainly accurate” to “certainly inaccurate,” in which researchers utilized that information, in addition to emotions and anxiety levels, to build a model and initiate 15 categories.
After that, all the information was examined and judged by a Georgia Tech computer. Most of the time, about 65 percent, was on point with the opinions of humans, regarding what tweets were deemed credible.
“Tweets with booster words, such as ‘undeniable,’ and positive emotion terms, such as ‘eager’ and ‘terrific,’ were viewed as highly credible,” Mitra stated.
“Words indicating positive sentiment but mocking the impracticality of the event, such as ‘ha,’ ‘grins’ or ‘joking,’ were seen as less credible. So were hedge words, including ‘certain level’ and ‘suspects,’” she continued.
Surprisingly, posts with high number of retweets received lower credibility scores, while replies on longer message tweets were deemed as more credible, based on the findings.
“When combined with other signals, such as event topics or structural information, our linguistic result could be an important building block of an automated system,” researchers concluded.