AI-DRIVEN E-COMMERCE REVIEW ANALYSIS USING BERT ALGORITHM

This project is designed to provide users with meaningful insights from customer feedback by analyzing reviews collected from popular e-commerce platforms like Amazon and Flipkart. It allows users to input product URLs, from which reviews are automatically extracted using robust web scraping techniques. These techniques are implemented using tools like Selenium and BeautifulSoup to handle dynamic content and retrieve multiple pages of reviews effectively. Once the reviews are collected, the backend processes those using advanced Natural Language Processing (NLP) techniques. The primary goals of this analysis are to perform sentiment classification, extract important keyphrases, and generate concise summaries of user opinions. These NLP tasks are powered by powerful Python libraries such as Hugging Face Transformers, Spacy, and Torch. BERT-based models help in understanding the contextual meaning of the text, while summarization is handled using pretrained models like BART. This enables the system to identify both positive and negative aspects of a product, such as “The phone has a great camera but poor battery life.” The application is built with a modern tech stack. The frontend is developed using React.js, ensuring a responsive and user- friendly interface. The backend ais powered by Node.js, which handles URL input, scraping logic, and connections with the NLP modules. All processed data, including raw reviews, sentiment labels, keyphrases, and summaries, are stored in a MongoDB database for easy access and retrieval. By combining web scraping and AI-powered text analysis, the project offers a smart solution for buyers to evaluate products based on real customer experiences. Instead of going through hundreds of reviews manually, users can now get a concise, clear summary that highlights important features and issues. This project demonstrates the effective use of machine learning and web technologies in solving real-world problems.