Countering hate on social media: Large scale classification of hate and counter speech

Garland, Joshua; Ghazi-Zahedi, Keyan; Young, Jean-Gabriel; Galešić, Mirta

doi:10.18653/v1/2020.alw-1.13

Cited by 39 publications

(36 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Observing civil interactions might even reduce the impact of incivility by breaking 'hate norms.' Evidence supporting this expectation comes from research showing that counter-speech can re-civilise online discourses (Garland, Ghazi-Zahedi, Young, Hébert-Dufresne, & Galesic, 2020;Ziegele, Jost, Frieß, & Naab, 2019). Overall, we formulated the following two hypotheses: H6.…”

Section: The Perceived Environmentmentioning

confidence: 99%

Roots of Incivility: How Personality, Media Use, and Online Experiences Shape Uncivil Participation

Frischlich

Schatto‐Eckrodt

Boberg

et al. 2021

MaC

View full text Add to dashboard Cite

Online media offer unprecedented access to digital public spheres, largely enhancing users’ opportunities for participation and providing new means for strengthening democratic discourse. At the same time, the last decades have demonstrated that online discourses are often characterised by so-called ‘dark participation’ the spreading of lies and incivility. Using ‘problematic behaviour theory’ as framework and focusing on incivility as a specific form of dark participation, this article investigates the role of users’ personal characteristics, media use, and online experiences in relation to offensive and hateful online behaviour. Using a random-quota survey of the German population, we explored how dark personality traits, political attitudes and emotions, the frequency and spaces of online-media use, and users’ experiences with both civil and uncivil online discourses predicted participants own uncivil behaviour, such as posting, sharing, or liking uncivil content. We found that 46% of the participants who had witnessed incivility in the last three months also engaged in uncivil participation. A hierarchical logistic regression analysis showed that incivility was associated with manipulative personality traits as measured by the dark triad, right-wing populist voting intentions, and frequent social-media use. Experiences with both civil comments and hate speech predicted higher levels of uncivil participation. The strongest predictor was participants’ personal experiences with online victimisation. Overall, the results confirmed that dark participation in the sense of uncivil engagement results from the interplay of personality traits, an online environment that allows for deviant engagement, and, most importantly, participants’ experiences in said environment.

show abstract

Section: The Perceived Environmentmentioning

confidence: 99%

Roots of Incivility: How Personality, Media Use, and Online Experiences Shape Uncivil Participation

Frischlich

Schatto‐Eckrodt

Boberg

et al. 2021

MaC

View full text Add to dashboard Cite

show abstract

“…Different machine learning approaches have been applied with varied success, including but not limited to support vector machines and random forests to convolutional and recurrent neural networks (Zhang and Luo, 2019;Bosco et al, 2018;de Gibert et al, 2018b;Kshirsagar et al, 2018;Malmasi and Zampieri, 2018;Pitsilis et al, 2018;Al-Hassan and Al-Dossari, 2019;Vidgen and Yasseri, 2020;Zimmerman et al, 2018). More recently, Garland et al (2020) used an ensemble learning algorithm to classify both hate speech and counter speech in a curated collection of German messages on Twitter. Unfortunately, these approaches require labeled sets of speech to train classifiers and therefore risk not transferring from one type of harmful speech (e.g.…”

Section: Previous Workmentioning

confidence: 99%

When the Echo Chamber Shatters: Examining the Use of Community-Specific Language Post-Subreddit Ban

Trujillo¹,

Rosenblatt²,

Anda-Jáuregui³

et al. 2021

Proceedings of the 5th Workshop on Online Abuse and Harms (WOAH 2021)

Self Cite

View full text Add to dashboard Cite

Community-level bans are a common tool against groups that enable online harassment and harmful speech. Unfortunately, the efficacy of community bans has only been partially studied and with mixed results. Here, we provide a flexible unsupervised methodology to identify in-group language and track user activity on Reddit both before and after the ban of a community (subreddit). We use a simple word frequency divergence to identify uncommon words overrepresented in a given community, not as a proxy for harmful speech but as a linguistic signature of the community. We apply our method to 15 banned subreddits, and find that community response is heterogeneous between subreddits and between users of a subreddit. Top users were more likely to become less active overall, while random users often reduced use of in-group language without decreasing activity. Finally, we find some evidence that the effectiveness of bans aligns with the content of a community. Users of dark humor communities were largely unaffected by bans while users of communities organized around white supremacy and fascism were the most affected. Altogether, our results show that bans do not affect all groups or users equally, and pave the way to understanding the effect of bans across communities.

show abstract

“…Automatic detection of abusive language can help identify and report harmful accounts and acts, and allows counter narratives (Chung et al, 2019;Garland et al, 2020;Ziems et al, 2020). Due to the volume of online text and the mental impact on humans who are employed to moderate online abusive language -moderators of abusive online content have been shown to develop serious PTSD and depressive symptoms (Casey Newton, 2020)it is urgent to develop systems to automate the detection and moderation of online abusive language.…”

Section: Introductionmentioning

confidence: 99%

Annotating Online Misogyny

Zeinert¹,

Inie²,

Derczynski³

2021

Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Confer

View full text Add to dashboard Cite

Online misogyny, a category of online abusive language, has serious and harmful social consequences. Automatic detection of misogynistic language online, while imperative, poses complicated challenges to both data gathering, data annotation, and bias mitigation, as this type of data is linguistically complex and diverse. This paper makes three contributions in this area: Firstly, we describe the detailed design of our iterative annotation process and codebook. Secondly, we present a comprehensive taxonomy of labels for annotating misogyny in natural written language, and finally, we introduce a high-quality dataset of annotated posts sampled from social media posts.

show abstract

Countering hate on social media: Large scale classification of hate and counter speech

Cited by 39 publications

References 30 publications

Roots of Incivility: How Personality, Media Use, and Online Experiences Shape Uncivil Participation

Roots of Incivility: How Personality, Media Use, and Online Experiences Shape Uncivil Participation

When the Echo Chamber Shatters: Examining the Use of Community-Specific Language Post-Subreddit Ban

Annotating Online Misogyny

Contact Info

Product

Resources

About