HateXplain: A Benchmark Dataset for Explainable Hate Speech Detection

TL;DR

HateXplain introduces a rationale-annotated benchmark for hate speech detection, collecting ~20,000 posts from Twitter and Gab with three labels per item: hate/offensive/normal, target communities, and human-identified rationales. The authors evaluate multiple models, including CNN-GRU, BiRNN variants, and BERT, under standard and rationale-supervised training, demonstrating that incorporating human rationales can improve performance and reduce bias, though explainability metrics vary by model and setup. They define a comprehensive evaluation framework inspired by ERASER, including three bias metrics (Subgroup, BPSN, BNSP) and ground-truth–driven plausibility/faithfulness measures, enabling a holistic assessment of explainability and fairness. The work highlights the importance of explanations in hate speech detection and provides a public dataset and tools to advance research toward more interpretable and less biased models in this high-stakes domain.

Abstract

Hate speech is a challenging issue plaguing the online social media. While better models for hate speech detection are continuously being developed, there is little research on the bias and interpretability aspects of hate speech. In this paper, we introduce HateXplain, the first benchmark hate speech dataset covering multiple aspects of the issue. Each post in our dataset is annotated from three different perspectives: the basic, commonly used 3-class classification (i.e., hate, offensive or normal), the target community (i.e., the community that has been the victim of hate speech/offensive speech in the post), and the rationales, i.e., the portions of the post on which their labelling decision (as hate, offensive or normal) is based. We utilize existing state-of-the-art models and observe that even models that perform very well in classification do not score high on explainability metrics like model plausibility and faithfulness. We also observe that models, which utilize the human rationales for training, perform better in reducing unintended bias towards target communities. We have made our code and dataset public at https://github.com/punyajoy/HateXplain

Figures (7)

• Figure 1: Ground truth attention.
• Figure 2: Representation of the general model architecture showing how the attention of the model is trained using the ground truth (GT) attention. $\lambda$ controls how much effect the attention loss has on the total loss.
• Figure 3: Community-wise results for each of the bias metrics.
• Figure 4: The classification interface. The annotator is provided with 20 text messages and asked to selected the correct type and target of the message.
• Figure 5: Rationale highlight. The annotators are asked to highlight the portions of the text that would justify the label.