Veracity and Velocity of Social Media Content during Breaking News: Analysis of November 2015 Paris Shootings

Social media sources are becoming increasingly important in journalism. Under breaking news deadlines semi-automated support for identification and verification of content is critical. We describe a large scale content-level analysis of over 6 million Twitter, You Tube and Instagram records covering the first 6 hours of the November 2015 Paris shootings. We ground our analysis by tracing how 5 ground truth images used in actual news reports went viral. We look at velocity of newsworthy content and its veracity with regards trusted source attribution. We also examine temporal segmentation combined with statistical frequency counters to identify likely eyewitness content for input to real-time breaking content feeds. Our results suggest attribution to trusted sources might be a good indicator of content veracity, and that temporal segmentation coupled with frequency statistical metrics could be used to highlight in real-time eyewitness content if applied with some additional text filters.