2017 SIAM Conference on Computational Science and Engineering
Abstract. We compare and evaluate the effectiveness of Support Vector Machine and Convolutional Neural Network models on hate speech identification tasks given a 15K tweet data set. We determine refinements and next steps required to produce a methodology for analyzing the social discourse patterns of those who engage in and are the victims of hate speech online, particularly in the context of packs of trolls engaged in collective action. We hypothesize that a better understanding of these dynamics will allow us to propose and test potential mitigations to these events, which cause real harm to individuals, and threaten the promise of open discourse on the internet.
- Chrys M. Watson Ross, University of New Mexico, USA, firstname.lastname@example.org