Listening to Affected Communities to Define Extreme Speech: Dataset and Experiments
This work addresses the challenge of accurately identifying hate speech for marginalized communities, offering an inclusive approach to dataset creation that could improve content moderation.
The paper tackles the problem of defining and detecting hate speech by introducing XTREMESPEECH, a dataset of 20,297 social media passages from four countries, created through direct involvement of affected communities, which results in more representative data and facilitates the removal of harmful content.
Building on current work on multilingual hate speech (e.g., Ousidhoum et al. (2019)) and hate speech reduction (e.g., Sap et al. (2020)), we present XTREMESPEECH, a new hate speech dataset containing 20,297 social media passages from Brazil, Germany, India and Kenya. The key novelty is that we directly involve the affected communities in collecting and annotating the data - as opposed to giving companies and governments control over defining and combatting hate speech. This inclusive approach results in datasets more representative of actually occurring online speech and is likely to facilitate the removal of the social media content that marginalized communities view as causing the most harm. Based on XTREMESPEECH, we establish novel tasks with accompanying baselines, provide evidence that cross-country training is generally not feasible due to cultural differences between countries and perform an interpretability analysis of BERT's predictions.