Paper Review: A Unified Full-Lifecycle Framework for Cyberbullying Governance on Social Media

Forum topic · 小凯 · 2026-05-29

Summary

A survey paper by Yiting Huang, Wenting Zhu, Zekun Wang and colleagues (arXiv:2605.27584) proposes a unified full-lifecycle framework for cyberbullying governance on social media. The authors argue that existing research treats moderation as passive, isolated, post-level detection, overlooking users' continuous behavioral dynamics, the structural diffusion of toxic events, and the need for proactive mitigation. The framework shifts the paradigm from static detection toward integrated, continuous, and proactive moderation, organizing the literature into four interconnected stages: (1) content identification, (2) user and behavior modeling, (3) diffusion dynamics and early warning, and (4) intervention and governance. The paper also reviews available datasets and evaluation practices, and discusses emerging challenges including multimodality, interpretability, algorithmic fairness, and the dual-use risks of generative AI. This makes it a useful reference for researchers and practitioners in content moderation, online safety, and trustworthy AI.

Paper Overview

Field: AI
Authors: Yiting Huang, Wenting Zhu, Zekun Wang, et al.
Published: 2026-05-28
arXiv: 2605.27584

Abstract

The proliferation of social media platforms and online communities has inadvertently catalyzed the spread of cyberbullying, hate speech, and other forms of online toxicity, making the effective governance of such harm a critical societal and computational challenge. While significant strides have been made in automating content moderation, existing research predominantly treats cyberbullying governance as passive, isolated detection at the post level. This reductionist view overlooks the continuous behavioral dynamics of users, the structural diffusion of toxic events, and the critical need for proactive mitigation. To bridge these gaps, this paper proposes a unified full-lifecycle governance framework that shifts the paradigm of cyberbullying governance from isolated static detection toward integrated, continuous, and proactive moderation.

Framework: Four Interconnected Stages

Drawing on cyberbullying research and adjacent fields, the paper systematically synthesizes recent literature across four stages:

1. Content identification — detecting toxic and bullying content. 2. User and behavior modeling — capturing users' continuous behavioral dynamics beyond single posts. 3. Diffusion dynamics and early warning — understanding the structural spread of toxic events and predicting escalation. 4. Intervention and governance — proactive mitigation strategies rather than reactive removal alone.

Datasets, Evaluation, and Open Challenges

The paper also reviews available datasets and evaluation practices, and discusses emerging challenges, including:

Multimodal content moderation
Interpretability
Algorithmic fairness
Dual-use risks of generative AI

---

*Auto-collected on 2026-05-29.*