Part-of-Speech Tagging Bahasa Jawa Menggunakan Model Pre-Trained Bidirectional Encoder Representation from Transformers

Ahmad Izzuddin (Universitas Panca Marga)
Nuzul Hikmah (Universitas Panca Marga)
Muhammad Alvin Ajry (Panca Marga University)

Journal of Information System
Vol 11 No 1 (2026): (Desember 2025 - Mei 2026)

Part-of-Speech Tagging (POS tagging) is the process of determining word classes in a text that is important in natural language processing. In Javanese, POS tagging is still a challenge due to limited linguistic resources and the complexity of the language. With the development of deep learning technology, the BERT (Bidirectional Encoder Representations from Transformers) fine-tuning method has been applied to classify word classes in Javanese, which is a language with limited resources. The javanese-bert-small model was trained using the UD_Javanese-CSUI dataset, and evaluated using precision, recall, F1-score, and accuracy metrics. The results showed that the model achieved good performance with an accuracy of 88,87%, and showed stability during training without significant overfitting. These findings indicate that the BERT-based approach is effective in handling word class ambiguity in Javanese and can be a stepping stone for further development in NLP systems for regional languages.