What causes a person to radicalize?
This was the subject of a fascinating talk delivered by Tamar Mitts, an assistant professor of international and public affairs at Columbia University, at a “data science day” hosted by the school on Wednesday. Mitts studied the efficacy of Twitter-disseminated propaganda supporting the self-identified Islamic State, or ISIS, in 2015 and 2016. To avoid the “obvious ethical issues” which attend to subjecting humans analysts to ISIS propaganda, Mitts said she used machine learning algorithms to identify and sort messages and videos into various categories, such as whether they contained violence. Then she parsed her dataset to uncover trends.