Minority (math) report: Scientists say algorithm can predict ISIS attacks
A team of computer scientists says they have created an algorithm that can decode patterns in the social media activity of Islamic State supporters to help predict when and where terrorist attacks are likely to occur.
In an article published Friday in the journal Science, researchers described their method of using an algorithm to sift through Russian social network VKontakte and look for pro-ISIS posts in multiple languages.
Social media has been a key tool for ISIS to disseminate its message, radicalize potential sympathizers and coordinate their activities.
Researchers chose the Russian network because of its linguistic and cultural diversity, as well as the fact that ISIS-supportive posts aren’t removed as quickly as they are on Facebook. VKontakte is the largest European social networking service, with over 90 million monthly users and 5 billion messages a day.
The study observed around 108,000 users in 196 informal, ad-hoc social pro-ISIS groups, which researchers referred to as “aggregates.” These groups would organically come together to discuss anything from tips for surviving drone strikes to declarations of their extremism.
The resulting data allowed the team, led by physicist Neil Johnson at the University of Miami, to create a “statistical model aimed at identifying behavioral patterns among online supporters of ISIS and used this information to predict the onset of major violent events.”
While law enforcement and intelligence agencies tend to focus on following the activities specific individuals, the team said that they focused on aggregates and studied how they came together and proliferated prior to an ISIS-related incident in the real world.
“It was like watching crystals forming. We were able to see how people were materializing around certain social groups; they were discussing and sharing information – all in real-time,” Johnson said in a statement. “The question is: Can there be a signal of how people are coming collectively together to do something without a proper system in place?”
There isn’t a conclusive answer to this question, but researchers were retroactively able to find at least one attack took place after a formation of these aggregates, the 2014 attack on the Syrian town of Kobane.
Fighting #ISIS With an Algorithm, University of Miami Physicists Try to Predict Attacks https://t.co/N6e09UolF7pic.twitter.com/xJFf6YsjCN
— University of Miami (@univmiami) June 17, 2016
“So the message is – find the aggregates, or at least a representative portion of them, and you have your hand on the pulse of the entire organization, in a way that you never could if you were to sift through the millions of internet users and track specific individuals, or specific hashtags,” Johnson said.
Researchers said that law enforcement and security forces could use their method to focus on a few groups to monitor for an impending attack.
They also said that the method could be used to track individuals who might launch lone-wolf attacks, such as Omar Mateen, who killed 49 people at a gay nightclub in Orlando, Florida on Sunday. Mateen, who pledged allegiance to ISIS, is believed to have been radicalized by the group online.
“With time, we would be able to track the trajectories of individuals through this ecology of aggregates,” Johnson said.
