Benchmarking Logistic Regression, SVM, Naive Bayes, and IndoBERT Fine-Tuning for Sentiment Analysis on Indonesian Product Reviews

Nabila Zakiyah Zahra, Salwa Farhanatussaidah, Nasywa Nur Afifah, Luluk Muthoharoh, Ardika Satria, Martin C.T. Manullang · 2026

Read Benchmarking Logistic Regression, SVM, Naive Bayes, and IndoBERT Fine-Tuning for Sentiment Analysis on Indonesian Product Reviews by Nabila Zakiyah Zahra, Salwa Farhanatussaidah, Nasywa Nur Afifah, Luluk Muthoharoh, Ardika Satria, Martin C.T. Manullang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-03439.txt.

Description

Nabila Zakiyah Zahra, Salwa Farhanatussaidah, Nasywa Nur Afifah, Luluk Muthoharoh, Ardika Satria, and Martin C. T. Manullang benchmark classic machine learning against IndoBERT for sentiment analysis of Indonesian product reviews. Tokopedia reviews offer a large, skewed three-class task where satisfaction signals matter to sellers at scale.

With word-frequency features, logistic regression, linear SVM, and naive Bayes served as baselines, while a fine-tuned IndoBERT used a custom weighted loss for imbalance. The linear SVM won clearly on accuracy and macro F1 in their setup, a gap the authors trace mainly to sampling: baselines saw the full corpus, the transformer only a subset. A Gradio app deploys the final model.

Catalog record

CREATORNabila Zakiyah Zahra, Salwa Farhanatussaidah, Nasywa Nur Afifah, Luluk Muthoharoh, Ardika Satria, Martin C.T. Manullang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE27 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.03439

Browse the full library · About the Tower of Babel