Accession Number : AD1053241


Title :   Understanding Violence Through Social Media


Descriptive Note : Technical Report


Corporate Author : Naval Postgraduate School Monterey United States


Personal Author(s) : Frost,Harold G ; Evans,Anthony W ; Hodges,Robert Jr H


Full Text : https://apps.dtic.mil/dtic/tr/fulltext/u2/1053241.pdf


Report Date : 01 Dec 2017


Pagination or Media Count : 65


Abstract : While social media analysis has been widely utilized to predict various market and political trends, its utilization to improve geospatial conflict prediction in contested environments remains understudied. To determine the feasibility of social media utilization in conflict prediction, we compared historical conflict data and social media metadata, utilizing over 829,537 geo-referenced messages sent through the Twitter network within Iraq from August 2013 to July 2014. From our research, we conclude that social media metadata has a positive impact on conflict prediction when compared with historical conflict data. Additionally, we find that utilizing the most extreme negative terminology from a locally derived social media lexicon provided the most significant predictive accuracy for determining areas that would experience subsequent violence. We suggest future research projects center on improving the conflict prediction capability of social media data and include social media analysis in operational assessments.


Descriptors :   social media , Iraq , regression analysis , violence , metadata , predictions , accuracy , online communications


Subject Categories : Information Science
      Sociology and Law
      Statistics and Probability


Distribution Statement : APPROVED FOR PUBLIC RELEASE