New algorithm predicts Daesh attacks

By Barry Eitel

SAN FRANCISCO

Researchers have created an algorithm they say accurately predicts Daesh attacks in the real world based off data gleaned from social media, scientists announced Friday.

Based in Florida, the scientists announced the algorithm less than a week after a deadly shooting at a gay nightclub in Orlando, Florida.

The shooting, considered by many to be the deadliest in U.S. history, killed 49 victims. Although the alleged shooter, Omar Mateen, pledged allegiance to Daesh, law enforcement agencies have not found any direct link between Mateen and the extremist group.

Researchers behind the algorithm, working at the University of Miami, hope to help law enforcement catch would-be attackers by modeling how extremist social media networks behave before anyone is killed.

While the algorithm now monitors groups instead of individuals, researchers believe it will soon be accurate enough to predict attacks similar to Orlando, where Daesh inspires an individual behind the violence but does not make direct contact.

The research was published Friday in the journal Science. The team monitored nearly 200 pro-Daesh online groups and found how those groups were able to grow, reincarnate if shut down and inspire attackers, like Mateen, with no known history of extremism.

“It was like watching crystals forming,” explained study author Neil Johnson in a statement. “We were able to see how people were materializing around certain social groups; they were discussing and sharing information—all in real-time.”

Johnson and his team adapted an equation used in chemistry to predict the development and growth of pro-Daesh networks.

Most importantly, the algorithm detects the behaviors and connections of these networks that foretell if online extremism will carry over into the real world.

“This removes the guess work,” Johnson said. “With that roadmap, law enforcement can better navigate what is going on, who is doing what, while state security agencies can better monitor what might be developing,” he added.