Washington: Scientists have found a way to identify extremists associated with terror groups like ISIS on social media, even before they post threatening content on their accounts.
The number and size of online extremist groups using social networks to harass users, recruit new members, and incite violence is rapidly increasing.
While social media platforms are working to combat this (in 2016, Twitter reported it had shut down 3,60,000 ISIS accounts) they traditionally rely heavily on users' reports to identify these accounts.
Once an account has been suspended, there is little that can be done to prevent a user from opening up a new account, or multiple accounts.
"Social media has become a powerful platform for extremist groups, ranging from ISIS to white nationalist 'alt-right' groups," said Tauhid Zaman of the Massachusetts Institute of Technology (MIT).
"These groups use social networks to spread hateful propaganda and incite violence and terror attacks, making them a threat to the general public," said Zaman.
The researchers collected Twitter data from approximately 5,000 'seed' users who were either known ISIS members or who were connected to many known ISIS members as friends or followers.
They obtained their names through news stories, blogs, and reports released by law enforcement agencies and think tanks.
In addition to reviewing the content of 4.8 million tweets from these users' timelines, including text, links, hash tags, and mentions, they also tracked account suspensions, as well as any suspensions of their friends and followers accounts.
Researchers focused on the account networks forged by known ISIS and Al Qaeda sympathisers and known foreign fighters and content that had been flagged by Twitter as terrorist in nature.
Using statistical modeling of extremist behaviour with optimised search policies and actual ISIS user data, the researchers developed a method to predict new extremist users, identify if more than one account belongs to the same user, as well as predict network connections of suspended extremist users who start a new account.
By tracking and comparing data on screen names, user name, profile images and banner images, the researchers were also able to identify 70 per cent of additional Twitter profiles held by extremist users, with only a two per cent incidence of misclassifying profiles.
"We created a new set of operational capabilities to deal with the threat posed by online extremists in social networks," said Lieutenant Colonel Christopher E Marks, of the US Army.
"We are able to predict who is an extremist before they post any content, and then able to predict where they will re-enter the network after they are suspended," Marks said.
"In short, we can automatically figure out who is an extremist and keep them off the social network," he said.
While the study was conducted using data from accounts belonging to ISIS extremists on Twitter, their methodology can be applied to any extremist group and any social network, researchers said.