Topic Evolution and Prediction on E-commerce Using Hybrid BERTopic-LDA-RF Model

The dataset chosen for this study is a collection of 25,109 computer reviews sourced from Jingdong Mall (JD.com), a popular Chinese e-commerce platform. These reviews focus specifically on computers and mobile, providing comprehensive feedback on product performance, user satisfaction, and technical issues. The dataset was gathered through web scraping, followed by translation of the Chinese reviews into English to facilitate analysis. Each review has been processed to ensure the quality and consistency of the data, enabling effective sentiment analysis and topic modeling. The dataset offers valuable insights into customer experiences and preferences, particularly for sentiment trends and product-specific feedback, making it ideal for the study’s purpose.