Twitter Inc (NYSE:TWTR) and Facebook Inc (NASDAQ:FB) have had to battle fake accounts. In the case of Twitter, robots control some of those fake accounts, and it can be pretty hard to tell sometimes. That’s about to change though, as a new analysis tool will make it easy to spot which accounts are being run by a social bot, a.k.a., some sort of automated software application.
BotOrNot analyzes Twitter accounts
According to Phys.org, the tool is called BotOrNot, and it looks over more than a thousand features from a Twitter Inc (NYSE:TWTR)’s social network, including their Twitter posts and other information. The tool provides the analysis in real time and assigns a probability on whether or not a particular Twitter account is being run by a bot.
Interestingly enough, the U.S. military and the National Science Foundation funded the research for developing the tool because they recognize that higher amounts of information flow have not only changed the way people communication, but also changed the way incorrect information spreads.
How the Twitter tool works
One of the researchers on the project, Alessandro Flammini, said they use a “statistical learning framework,” although the “secret sauce” is a number of features the tool users to tell the difference between social bots and human Twitter Inc (NYSE:TWTR) users. The researcher said the tool looks at the content and also the timing of the Twitter account’s tweets. The tool also takes into account the structure of the accounts’ networks.
To develop the tool, researchers looked at the habits of Twitter bots created by a Texas A&M University professor’s infolab. Then they “train statistical models” so that they can tell the difference between humans and social bots. Researchers say the system is very accurate. Using the AUROC evaluation measure, BotOrNot gets a .95 score, with 1.0 indicating perfect accuracy.
How many bots are on Twitter?
Fil Menczer, another researcher at IU, said they wanted to know how many bots are trolling the waters on Twitter Inc (NYSE:TWTR). He said they wanted to know just how bad the bot problem is on the social network, whether the number of bots adds up to the hundreds, thousands or even millions. He said they know there are “lots of bots” on there and that many of them “are totally benign.” However, he also said they found some “nasty bots used to mislead, exploit and manipulate discourse with rumors, spam, malware, misinformation, political astro-turf and slander.”
Both of the researchers believe that such bots could be dangerous, potentially creating panic when there’s an emergency, affect the stock markets, make it easier for cyber-criminals to thrive, and hold back public policy. They said their goal is to provide truthful information to counter all the misinformation that’s out there.
Speaking of misinformation, Facebook Inc (NASDAQ:FB) got into a lot of trouble with its Related Articles feature lately. It sure sounds like they could use some help from the folks at IU.