Skip to content
SDGs Center UDINUS

Evidence record · UDINUS · SDGs

IndoBERT-based Indonesian Cyberbullying Detection with Multi-stage Labeling

Rektorat UDINUS 2024 Published 2026-10-30

Evidence record · UDINUS · SDGs

Unit
Rektorat UDINUS
Period
2024
Goals
Record
acb58dd9-03a0-4fc2-83a7-28cd44ef51cf
Published
2026-10-30

Universitas Dian Nuswantoro conducted a study addressing cyberbullying on social media in Indonesia and the need to protect users from irresponsible behavior. Researchers Novandian, Yohanes Deny; Luthfiarta, Ardytha; Assyifa, Dhiaka Shabrina; Setiawan, Johanes; Cahyaningrum, Lailatul; Althoff, Noval; Rahayu, Mufida; Nugraha, Adhitya; and Rismiyati participated in the activity. The study focused on developing an efficient and effective text classification model to detect indications of bullying.

The activity collected a dataset from Twitter by searching for keywords related to body shaming. The dataset was labeled using Large Language Model and Vader Lexicon approaches for more accurate labeling. Because the labeled dataset was imbalanced, SMOTE was applied for oversampling to obtain a balanced dataset. IndoBERT was used as the detection model, and fine-tuning resulted in 96.7% accuracy.

The proposed model was examined using real data from a survey related to bullying sentences conducted with 20 students from Dian Nuswantoro University. The results showed that the models accurately predicted that the sentences contained elements of bullying. This activity presents institutional research by Universitas Dian Nuswantoro on cyberbullying detection, using text classification, dataset labeling, oversampling, and model examination to address irresponsible behavior on social media.

Reference https://www.scopus.com/pages/publications/85213370192?origin=resultslist