Satirists, Supporters and the Real Deal: Identifying Extremists in a Sea of Social Media

No ratings

Presented at CounterMeasure 2014 by

Justin Seitz

With the widespread adoption of social media, in particular Twitter, by extremists around the globe there is increasing interest in how to identify and monitor them. Traditional approaches include keyword monitoring, text analytics or sentiment analysis, but in many cases these techniques are really only useful once the target group has already been identified in the first place. This talk will demonstrate a different approach, via automated image analysis, that can be used to rapidly establish a dataset to begin monitoring extremist groups on Twitter. The issues that go along with image vs. text analysis will also be discussed, including the challenges of creating a classifier for the computer vision non-expert. Other technical topics will include solving six degrees of separation, graph exploration and how to stand up a Twitter intelligence gathering and analysis system in less than 30 minutes and for zero cost. Expect some Python.