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

Cyberbullying (CB) presents a pressing threat, especially to children, underscoring the urgent need for robust detection systems to ensure online safety. However, progress in developing such systems is hindered by the scarcity of large, labeled datasets that are specifically tailored for specialized tasks and the target age groups. Creating these datasets relies heavily on human annotation, which not only strains resources but also raises significant ethical and legal concerns due to annotators' exposure to harmful content, notwithstanding the acquisition of this type of data from vulnerable populations such as children. In this paper, we address these challenges by leveraging Large Language Models (LLMs) to generate synthetic data and labels. Our experiments demonstrate that synthetic data enables BERT-based CB classifiers to achieve performance close to that of those trained on fully authentic datasets (75.8% vs. 81.5% accuracy). Additionally, LLMs can effectively label authentic yet unlabeled data, allowing BERT classifiers to attain a comparable performance level (79.1% vs. 81.5% accuracy). These results highlight the potential of LLMs as a scalable, ethical, and cost-effective solution for generating data for CB detection.

@article{kazemi2025_2502.15860,
  title={ Synthetic vs. Gold: The Role of LLM-Generated Labels and Data in Cyberbullying Detection },
  author={ Arefeh Kazemi and Sri Balaaji Natarajan Kalaivendan and Joachim Wagner and Hamza Qadeer and Brian Davis },
  journal={arXiv preprint arXiv:2502.15860},
  year={ 2025 }
}