Smart Review Analyzer: A Multiplatform AI System for Aspect-Based Sentiment Analysis on E-Commerce Products

Abstract

With the increasing popularity of user-generated reviews on the e-commerce domain, including the likes of Amazon, there has been increasing difficulty for consumers in processing huge amounts of feedback in making a buying decision. One very high fair average rating (such as 4.5 stars) may confuse some latent defects hidden within the comments of the individuals. The paper described a smart review analysis system based on Aspect-Based Sentiment Analysis (ABSA) that classifies customer feedback according to the main criteria of product quality, packaging, delivery, and price. At the very heart of the system is a fine-tuned DeBERTa model, trained on a custom-labeled dataset generated by automated web scraping using requests and BeautifulSoup. The system proposed goes beyond simple polarity determination by linking sentiments with the aspects of the analysis under consideration. The solution is implemented both as a web app (React) and a cross-platform mobile app (Flutter), with Firebase as the authentication and backend service provider. The platform enables consumers to look up real-time product sentiment data, while sellers gain interactive dashboards to visualize customer satisfaction, identify problems, and analyze product profitability. This study exhibits the role of AI-powered sentiment analysis in advancing transparency and decision-making for consumers and vendors in digital marketplaces.