Contextualizing Hate Speech Classifiers with Post-hoc Explanation
This addresses bias in hate speech detection for social media platforms, but it is incremental as it builds on existing BERT classifiers and explanation methods.
The paper tackled bias in hate speech classifiers that produce false positives for group identifiers like 'gay' or 'black' by proposing a regularization technique based on post-hoc explanations to encourage learning from context. The approach improved over baselines in reducing false positives on out-of-domain data while maintaining in-domain performance.
Hate speech classifiers trained on imbalanced datasets struggle to determine if group identifiers like "gay" or "black" are used in offensive or prejudiced ways. Such biases manifest in false positives when these identifiers are present, due to models' inability to learn the contexts which constitute a hateful usage of identifiers. We extract SOC post-hoc explanations from fine-tuned BERT classifiers to efficiently detect bias towards identity terms. Then, we propose a novel regularization technique based on these explanations that encourages models to learn from the context of group identifiers in addition to the identifiers themselves. Our approach improved over baselines in limiting false positives on out-of-domain data while maintaining or improving in-domain performance. Project page: https://inklab.usc.edu/contextualize-hate-speech/.