SynBullying: A Multi LLM Synthetic Conversational Dataset for Cyberbullying Detection

TL;DR

SynBullying is a synthetic multi-turn conversational dataset created using large language models to improve cyberbullying detection, offering realistic, context-aware, and fine-grained annotations for better analysis and model training.

Contribution

The paper introduces SynBullying, a novel synthetic dataset generated by LLMs that captures realistic, multi-turn cyberbullying interactions with detailed annotations for enhanced detection methods.

Findings

Effective in training cyberbullying detection models

Improves model performance when used as augmentation data

Captures diverse cyberbullying behaviors and linguistic patterns

Abstract

We introduce SynBullying, a synthetic multi-LLM conversational dataset for studying and detecting cyberbullying (CB). SynBullying provides a scalable and ethically safe alternative to human data collection by leveraging large language models (LLMs) to simulate realistic bullying interactions. The dataset offers (i) conversational structure, capturing multi-turn exchanges rather than isolated posts; (ii) context-aware annotations, where harmfulness is assessed within the conversational flow considering context, intent, and discourse dynamics; and (iii) fine-grained labeling, covering various CB categories for detailed linguistic and behavioral analysis. We evaluate SynBullying across five dimensions, including conversational structure, lexical patterns, sentiment/toxicity, role dynamics, harm intensity, and CB-type distribution. We further examine its utility by testing its performance as standalone training data and as an augmentation source for CB classification.