Cyberbullying Detection via Aggression-Enhanced Prompting

2025-08-08

AI Summary

This study addresses the limited generalization capability of large language models (LLMs) in detecting covert, context-dependent cyberbullying on social media. To tackle this, we propose a multi-task enhancement framework with aggression detection as an auxiliary task. Our core innovation is the *Aggression-Aware Prompting Pipeline*, which explicitly incorporates aggression prediction signals into the instruction-tuning prompt structure, enabling context-aware joint modeling while avoiding the instability inherent in conventional multi-task learning. The method integrates zero-shot/few-shot inference, LoRA-based efficient fine-tuning, and advanced prompt engineering, and is systematically evaluated across multiple mainstream cyberbullying benchmarks. Experimental results demonstrate consistent superiority over baseline LoRA fine-tuning, yielding average improvements of 3.2–5.8 percentage points in F1 and AUC scores. Notably, the approach significantly enhances robustness in identifying implicit bullying instances, including sarcasm and irony.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Multiagent Systems: Adversarial Agents

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

Detecting cyberbullying on social media remains a critical challenge due to its subtle and varied expressions. This study investigates whether integrating aggression detection as an auxiliary task within a unified training framework can enhance the generalisation and performance of large language models (LLMs) in cyberbullying detection. Experiments are conducted on five aggression datasets and one cyberbullying dataset using instruction-tuned LLMs. We evaluated multiple strategies: zero-shot, few-shot, independent LoRA fine-tuning, and multi-task learning (MTL). Given the inconsistent results of MTL, we propose an enriched prompt pipeline approach in which aggression predictions are embedded into cyberbullying detection prompts to provide contextual augmentation. Preliminary results show that the enriched prompt pipeline consistently outperforms standard LoRA fine-tuning, indicating that aggression-informed context significantly boosts cyberbullying detection. This study highlights the potential of auxiliary tasks, such as aggression detection, to improve the generalisation of LLMs for safety-critical applications on social networks.

Problem

• Enhancing cyberbullying detection via aggression context
• Improving LLM generalization with auxiliary tasks
• Evaluating prompt strategies for social media safety

Innovation

• Integrates aggression detection as auxiliary task
• Uses enriched prompt pipeline for context
• Leverages multi-task learning and LoRA