Students following last week’s presidential debate on Twitter might be unaware that their tweets were being analyzed by more than just their followers.
Christopher Healey, a computer science professor at N.C. State University, has developed a program — known as Twitter Sentiment — that monitors tweets in order to track voter opinions.
“We were looking for a way to measure and visualize texts,” he said, adding that Twitter’s popularity makes it a good place to track interest in politics.
The model searches for usage of more than 1,100 words and categorizes tweets on a range from “active” to “subdued” and from “pleasant” to “unpleasant.” Healey said the scales of emotion have been measured by behavioral psychologists.
Prior to the first presidential debate, Healey said the amount of tweets about Republican presidential nominee Mitt Romney and President Barack Obama were fairly even.
But at one point, tweets about Romney — who most political analysts said performed better in the debate — began outnumbering tweets about Obama two to one.
The program tracked a sample of 350,000 tweets out of 10.4 million total tweets about the debate.
But Healey admitted the application is not perfect. Tracking Twitter, which limits posts to 140 characters, can be problematic since messages might be misinterpreted.
“The trouble with things like Twitter is people don’t often use correct grammar or complete spelling,” he said.
