| Abstrakt: |
Proud Boys, a contemporary radical extremist group has exploited the convenience and reach of social media platforms to spread its hateful ideology. To combat this spread, this paper analyzes social media dialogue surrounding Proud Boys protests from multiple perspectives to understand: (i) the outlook and profiles of users who support and reject Proud Boys; (ii) the network structure to determine influencers and communities; and (iii) the degree of engagement of tweets and the activity of their authors. The analyses indicate that conservative, religious, right-wing segments of the populace support Proud Boys, and their support is rooted in patriotism and defense of American values. Socialists and progressives oppose their radical and hateful ideology. Proud Boys' topic network is fragmented with low density and is comprised of distinct, small communities. Authors of opposing tweets have a more extensive set of friends and followers, and their tweets receive more likes and retweets. Based on these results, it can be inferred that tweets endorsing Proud Boys' ideology do not appear to proliferate virally, but instead reach limited audiences through smaller clusters. These analyses and results then form the basis of a classification framework, which can accurately detect tweets supportive of Proud Boys with an accuracy of 0.84 and AUC ROC of 0.89. The capability of the framework to identify and demote supporting tweets offers an extra layer of mitigation that can be employed to stem their spread. The research showcases the promise of mining social media feeds to understand why and how radical extremism spreads and is a gateway for future researchers. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink