Hate speech detection: Challenges and solutions

MacAvaney, Sean; Yao, Hao Ren; Yang, Eugene; Russell, Katina; Goharian, Nazli; Frieder, Ophir

doi:10.1371/journal.pone.0221152

Cited by 362 publications

(266 citation statements)

References 23 publications

Supporting

Mentioning

238

Contrasting

Unclassified

Order By: Relevance

“…The field has been recently surveyed in [8,9]. The vast majority of the papers analyzed in [8] describes approaches to hate speech detection based on supervised learning, where the task is treated as a sentence-or message-level binary text classification task.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Time of Your Hate: The Challenge of Time in Hate Speech Detection on Social Media

et al. 2020

View full text Add to dashboard Cite

The availability of large annotated corpora from social media and the development of powerful classification approaches have contributed in an unprecedented way to tackle the challenge of monitoring users’ opinions and sentiments in online social platforms across time. Such linguistic data are strongly affected by events and topic discourse, and this aspect is crucial when detecting phenomena such as hate speech, especially from a diachronic perspective. We address this challenge by focusing on a real case study: the “Contro l’odio” platform for monitoring hate speech against immigrants in the Italian Twittersphere. We explored the temporal robustness of a BERT model for Italian (AlBERTo), the current benchmark on non-diachronic detection settings. We tested different training strategies to evaluate how the classification performance is affected by adding more data temporally distant from the test set and hence potentially different in terms of topic and language use. Our analysis points out the limits that a supervised classification model encounters on data that are heavily influenced by events. Our results show how AlBERTo is highly sensitive to the temporal distance of the fine-tuning set. However, with an adequate time window, the performance increases, while requiring less annotated data than a traditional classifier.

show abstract

Section: Related Workmentioning

confidence: 99%

“…Being computationally very expensive, researchers only recently succeeded in training BERT deep neural networks. The aforementioned survey [9] includes the more recent BERT model and introduces a modified and more transparent version of an SVM classifier that does not, however, outperform BERT.…”

Section: Related Workmentioning

confidence: 99%

Time of Your Hate: The Challenge of Time in Hate Speech Detection on Social Media

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Finally, to address the calls for openness [103], we are making the source code 24 of the classifiers public for further research and development and for software engineers to adopt in real systems and applications.…”

Section: Key Contributionsmentioning

confidence: 99%

Developing an online hate classifier for multiple social media platforms

Salminen

Hopf²,

Chowdhury

et al. 2020

Hum. Cent. Comput. Inf. Sci.

208

View full text Add to dashboard Cite

been identified as a major threat on online social media platforms. Pew Research Center [13] reports that among 4248 adults in the United States, 41% have personally experienced harassing behavior online, whereas 66% witnessed harassment directed towards others. Around 22% of adults have experienced offensive name-calling, purposeful embarrassment (22%), physical threats (10%), and sexual harassment (6%), among other types of harassment. Social media platforms are the most prominent grounds for such toxic behavior. Even though they often provide ways of flagging offensive and hateful content, only 17% of all adults have flagged harassing conversation, whereas only 12% of adults have reported someone for such acts [13].

show abstract

“…Emerging literature in the social and computational sciences increasingly adopts the view that interpretability of models is likewise important [53]. In certain cases, the complex features of state-of-the-art neural models may make it challenging for researchers to understand the predictions made.…”

Section: Hate Speech On Social Media: From Classification To Charactementioning

confidence: 99%

Bots and online hate during the COVID-19 pandemic: case studies in the United States and the Philippines

Uyheng

Carley

2020

J Comput Soc Sc

View full text Add to dashboard Cite

Online hate speech represents a serious problem exacerbated by the ongoing COVID-19 pandemic. Although often anchored in real-world social divisions, hate speech in cyberspace may also be fueled inorganically by inauthentic actors like social bots. This work presents and employs a methodological pipeline for assessing the links between hate speech and bot-driven activity through the lens of social cybersecurity. Using a combination of machine learning and network science tools, we empirically characterize Twitter conversations about the pandemic in the United States and the Philippines. Our integrated analysis reveals idiosyncratic relationships between bots and hate speech across datasets, highlighting different network dynamics of racially charged toxicity in the US and political conflicts in the Philippines. Most crucially, we discover that bot activity is linked to higher hate in both countries, especially in communities which are denser and more isolated from others. We discuss several insights for probing issues of online hate speech and coordinated disinformation, especially through a global approach to computational social science.

show abstract

Hate speech detection: Challenges and solutions

Cited by 362 publications

References 23 publications

Time of Your Hate: The Challenge of Time in Hate Speech Detection on Social Media

Time of Your Hate: The Challenge of Time in Hate Speech Detection on Social Media

Developing an online hate classifier for multiple social media platforms

Bots and online hate during the COVID-19 pandemic: case studies in the United States and the Philippines

Contact Info

Product

Resources

About