Professional · 2020-06 → 2021-02
IndoBERT Indonesian Text Corrector
Developed an Indonesian text corrector using IndoBERT transformer model to improve labeling accuracy for Government Sentiment Analysis on social media data.
PythonNLPIndoBERTSQLExcel
As part of Telkom Indonesia's data science division, I developed an Indonesian text corrector using IndoBERT — a transformer-based NLP model specifically trained on Indonesian language data.
The corrector was integrated into the Government Sentiment Analysis pipeline to improve labeling accuracy on social media data.
What I built
- Developed Indonesian text corrector using IndoBERT transformer model.
- Improved labeling accuracy for Government Sentiment Analysis on social media posts.
- Built and labeled datasets from government-related social media posts for model training.
- Directly improved downstream sentiment analysis model performance through quality data labeling.