Indonesian E-Commerce Fake and Authentic Reviews Dataset

This dataset contains 75,000 Indonesian e-commerce product reviews collected from Tokopedia and Shopee between January–June 2024, comprising 50,000 authentic verified reviews and 25,000 AI-generated fake reviews created using GPT-3.5-turbo. The data spans five product categories (electronics, fashion, home goods, beauty, and food) with balanced sentiment distribution (65% positive, 20% neutral, 15% negative). Authentic reviews were filtered for purchase verification and linguistic validity, while synthetic reviews were designed to mimic Indonesian consumer language patterns with slang, code-mixing, and cultural nuances. This dataset provides a realistic benchmark for developing and evaluating transformer-based adversarial models in fake review detection and digital trust research.