Synthetic vs. Gold: The Role of LLM-Generated Labels and Data in Cyberbullying Detection

2025-02-21

AI Summary

This work investigates whether synthetic data generated by large language models (LLMs) can effectively substitute for or augment real labeled data to improve cyberbullying detection. We systematically generate high-quality synthetic texts and labels using diverse state-of-the-art LLMs—including GPT, Llama, and Qwen—and train binary classifiers within standard NLP preprocessing and supervised learning pipelines. Results show that models trained solely on synthetic data achieve performance nearly matching that of real-data baselines; furthermore, hybrid training on real and synthetic data consistently improves accuracy by 2.1–3.8% across all three LLM families. This study provides the first empirical validation of LLM-generated synthetic data for cyberbullying detection, reveals critical sensitivity to LLM selection, and establishes synthetic data augmentation as a scalable, privacy-preserving, and ethically compliant paradigm for dataset enhancement.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Computer Vision: Large Vision Models

Application Category

• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• 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

Abstract

This study investigates the role of LLM-generated synthetic data in cyberbullying detection. We conduct a series of experiments where we replace some or all of the authentic data with synthetic data, or augment the authentic data with synthetic data. We find that synthetic cyberbullying data can be the basis for training a classifier for harm detection that reaches performance close to that of a classifier trained with authentic data. Combining authentic with synthetic data shows improvements over the baseline of training on authentic data alone for the test data for all three LLMs tried. These results highlight the viability of synthetic data as a scalable, ethically viable alternative in cyberbullying detection while emphasizing the critical impact of LLM selection on performance outcomes.

Problem

• LLM-generated synthetic data in cyberbullying
• Synthetic vs. authentic data performance
• Ethical scalability in cyberbullying detection

Innovation

• LLM-generated synthetic data
• augment authentic data
• ethical scalable alternative