Indonesian Shopee Product Review Sentiment Dataset for E-Commerce Sentiment Classification

This dataset contains 16,174 Indonesian-language Shopee product reviews collected for academic research on e-commerce sentiment classification. The reviews cover three product categories: Perawatan Diri, Elektronik, and Fashion. Sentiment labels were generated using rating-based rules: ratings 1-2 as Negatif, rating 3 as Netral, and ratings 4-5 as Positif. The dataset was used in a comparative experiment involving Support Vector Machine (SVM) with TF-IDF, fine-tuned IndoBERT, and GPT zero-shot classification. The dataset is highly imbalanced, with Positif reviews dominating the distribution, making it suitable for research on imbalanced sentiment classification in Indonesian e-commerce reviews. The uploaded files include the labeled dataset, data dictionary, dataset summary, README file, and citation metadata.