Tox-BART: Leveraging Toxicity Attributes for Explanation Generation of Implicit Hate Speech
This work addresses the challenge of content moderation by aiding moderators with explicit explanations, but it appears incremental as it builds on existing methods with specific improvements.
The paper tackled the problem of generating explanations for implicit hate speech by investigating the role of knowledge graph quality, finding that simpler models using toxicity signals outperform knowledge graph-infused models with performance variations such as +0.44 in BLEU on the SBIC dataset.
Employing language models to generate explanations for an incoming implicit hate post is an active area of research. The explanation is intended to make explicit the underlying stereotype and aid content moderators. The training often combines top-k relevant knowledge graph (KG) tuples to provide world knowledge and improve performance on standard metrics. Interestingly, our study presents conflicting evidence for the role of the quality of KG tuples in generating implicit explanations. Consequently, simpler models incorporating external toxicity signals outperform KG-infused models. Compared to the KG-based setup, we observe a comparable performance for SBIC (LatentHatred) datasets with a performance variation of +0.44 (+0.49), +1.83 (-1.56), and -4.59 (+0.77) in BLEU, ROUGE-L, and BERTScore. Further human evaluation and error analysis reveal that our proposed setup produces more precise explanations than zero-shot GPT-3.5, highlighting the intricate nature of the task.