Hungarian case study on automated detection of body-shaming comments using machine learning

Gőz, Franciska Noémi and B. Varga, Erika (2025) Hungarian case study on automated detection of body-shaming comments using machine learning. In: Proceedings of the International Conference on Formal Methods and Foundations of Artificial Intelligence. Eszterházy Károly Katolikus Egyetem Líceum Kiadó, Eger, pp. 65-77. ISBN 9789634963035

Preview

Text fmfai2025_pp065-077.pdf - Published Version Download (809kB) | Preview

Abstract

Social media facilitates online interactions but also enables bodyshaming comments which are often ambiguous. This paper presents a machine learning-based approach for detecting Hungarian body-shaming comments, an underrepresented area in NLP. A dataset of Facebook comments was collected and expanded with synthetic data. Using HuSpaCy and HuBERT, logistic regression and MLP classification models were trained with TF-IDF and SBERT embeddings. The best-performing model achieved 88% accuracy, demonstrating the potential of NLP techniques for moderating harmful online content in low-resource languages. The results highlight key challenges, including category overlap and class imbalance, emphasizing the need for context-aware classification methods in automated content moderation.

Actions (login required)

Edit Item