ISI Analysis of Jihadi Tweets Featured by MIT Technology Review

February 14, 2017

MIT Technology Review covered work by Adam Badawy and Emilio Ferrara that used data mining to analyze Twitter messages by 25,000 ISIS supporters. In the February 10, 2017 issue, MIT TR describes Badawy and Ferrara's exploration of 1.9 million tweets from now-closed accounts.

Among Badawy and Ferrara's conclusions: Violent rhetoric is deliberate to instill pride, excite prospective recruits and distinguish ISIS from the "spiritual hegemony" of its predecessors.

The pair focused solely on messages in Arabic, which they believe makes their work among the first to look at text in that language — rather than English — in depth. They first identified common words and phrases, or "stems", many of which related to violence, theological pronouncements and sectarianism. Badawy and Ferrara then used machine learning to classify the tweets by stem.

They discovered that more than 30 percent of the messages reference theology and violence, a distinct shift from al-Qaeda's previous, David-and-Goliath approach to Jihad. Say the authors, "ISIS, from the start, refused to portray itself as an underdog, focusing on its victories, atrocious violent acts against minorities ... and its call for an Islamic state."

They also demonstrated that violent rhetoric peaks after, and sometimes precedes, terrorist events.

Ferrara is an ISI research lead and USC research assistant professor. Badawy is a Ph.D. candidate, and The Rise of Jihadist Propaganda on Social Networks is his first paper. Of MIT TR's story, says Ferrara, "Not a shabby way to begin a Ph.D. publishing career!"