Tokopedia E Commerce Review In Indonesia

This dataset was obtained through web scraping using Python from customer reviews of a motorcycle oil product on Tokopedia in Indonesia, preserved in their original form, including typos, abbreviations, and informal language, to maintain linguistic authenticity. The research tests the hypothesis that reviews with overly polished language, generic praise lacking product-specific details, and perfect punctuation are more likely to be fake, whereas reviews containing detailed product references, informal expressions, spelling mistakes, and emotional tone are more likely to be genuine. The analysis shows that a small number of reviews appear potentially fake, typically consisting of short, generic praise unrelated to the product, while most reviews seem authentic and contain specific complaints such as unusual smell, abnormal oil thickness, leaking packaging, missing barcodes, and suspicions of counterfeit goods. These findings suggest that the majority of negative feedback likely reflects actual quality issues or the circulation of fake products, and can be used by e-commerce platforms to prioritize relevant reviews and develop linguistic-based fake review detection systems.