E-Commerce Marketplace Comment Dataset

The study employed a large-scale dataset collected in real-time from three major Indonesian e-commerce marketplaces—Tokopedia, Shopee, and Bukalapak—covering the period from January to June 2024. The dataset comprised 2,847,392 customer comments and reviews across 15,847 unique products and 3,241 sellers. Each record included text content, timestamps, seller responses and their timing, sales-rank changes, review helpfulness votes, and competitor sentiment trends. Approximately 98% of the comments were in Indonesian and 2% were mixed Indonesian-English. This comprehensive dataset provided the foundation for sentiment classification using IndoBERT, sentiment velocity calculation, and predictive modeling for early sales forecasting.