Hierarchical Detox
Title: Hierarchical Adversarial Correction to Mitigate Identity Term Bias in Toxicity Detection (2024)
Text: We correct the learning of toxicity detection by considering the mentioned targets at different levels of specificity.
Authors: Johannes Sch?fer, Ulrich Heid, Roman Klinger
- Paper: https://aclanthology.org/2024.wassa-1.4/
- Code: https://github.com/Johannes-Schaefer/hierarchical-detox