Confronting Abusive Language Online: A Survey from the Ethical and Human Rights Perspective

The pervasiveness of abusive content on the internet can lead to severe psychological and physical harm. Significant effort in Natural Language Processing (NLP) research has been devoted addressing this problem through detection related sub-areas, such as hate speech, toxicity, cyberbullying, etc. Although current technologies achieve high classification performance studies, it observed that real-life application technology cause unintended harms, silencing under-represented groups. We review a large body NLP automatic abuse with new focus ethical challenges, organized around eight established principles: privacy, accountability, safety security, transparency explainability, fairness non-discrimination, human control technology, professional responsibility, promotion values. In many cases, these principles relate not only situational codes, which may be context-dependent, but are fact connected universal rights, right freedom from discrimination, expression. highlight need examine broad social impacts bring rights considerations every stage life-cycle, task formulation dataset design, model training evaluation, deployment. Guided by principles, we identify several opportunities for rights-respecting, socio-technical solutions detect confront online abuse, including `nudging', `quarantining', value sensitive counter-narratives, style transfer, AI-driven public education applications.