Social media has become one of the biggest threats schools face in this day and age, and in 2014 we saw over 300 threats against schools posted on social networks, which is very alarming. As if detecting threats on Facebook, Twitter and other popular networks isn’t challenging enough, students are now using anonymous apps like Yik Yak and Burnbook to wreak havoc while attempting to shield their identities.
According to an article from the New York Times, “Since the app was introduced a little more than a year ago, it has been used to issue threats of mass violence on more than a dozen college campuses, including the University of North Carolina, Michigan State University and Penn State. Racist, homophobic and misogynist ‘yaks’ have generated controversy at many more, among them Clemson, Emory, Colgate and the University of Texas. At Kenyon College, a ‘yakker’ proposed a gang rape at the school’s women’s center.”
Given the perceived anonymity of posts on Yik Yak and Burnbook, our research indicates that individuals often make more direct and violent threats when compared to threats that have been made on Twitter and other social networks in the past. With no profile, friends or other dots that can be connected, schools and law enforcement agencies struggle to combat and prosecute those making anonymous threats.
Even worse, most schools have geofences that block the use of the most popular anonymous social network, Yik Yak, within a certain radius of their physical location, which means instead of having centralized conversations that can be closely monitored we have fractured conversations. Due to such, the difficulty of detecting threats exponentially increases based on a variety of factors (number of students, their distance from the school itself, etc.).
The fact is that linguistic patterns, namely stylistics, can play a large role in helping to identify those making anonymous threats by cross-referencing activity on social networks like Twitter to anonymous posts being made on Yik Yak or Burnbook. This is an area of research our company has focused heavily on, and we have been developing algorithms aimed at automating the process.