Embassies burning: toward a near-real-time assessment of social media using geo-temporal dynamic network analytics

Abstract

Effective crisis response requires rapid assessment of a situation in order to form actionable plans. Social media and traditional media are critical to this assessment. This paper describes a rapid ethnographic approach for extracting information from Twitter and news media and then assessing that information using dynamic network analysis techniques. Text mining high-dimensional network analytics and visualization are combined to provide an integrated approach to assessing large dynamic networks. This approach was used as the Benghazi consulate and the Egyptian embassy were attacked in 2012. This near-real-time assessment was set against a backdrop of ongoing data collection associated with the Arab Spring countries. This ongoing collection provided a baseline for Libya and Egypt against which the new data could be assessed. Herein, the outcome of that near-real-time assessment, the tools used, the lessons learned, and the results discovered are described. The same approach was used in other crisis events including SuperStorm Sandy, the Kenyan elections, from which examples are also drawn. We find that to be effective such analytics require the use of multiple media, deep dives into specific secondary issues, and a high-level assessment of not just who is doing what, but who is providing what information. Finally, we show the criticality of baseline data for interpreting the behavior during a crisis.