Shop Sense AI

Abstract

Online shopping is burdened with fragmented product data, unreliable reviews, and inefficient search tools. While big retailers and comparison sites have tried to solve this with basic aggregation, most of the current solutions lack intelligent analysis and real-time local inventory. In addition, this paper proposes a comprehensive AI-driven platform that leverages a multi-model artificial intelligence approach. This system utilizes a fine-tuned DistilBERT model to carry out advanced sentiment and aspect-based analysis of user reviews, filtering misinformation and generating a trustworthy credibility score. Simultaneously, this enables a strong visual search and product recognition from images through the YOLO object detection model. Driven by the intelligent search, this will be empowered by an NLP engine that is capable of handling complex queries in a conversational manner via text or voice. The proposed approach effectively creates a unified, intelligent, and trustable shopping assistant that consolidates the entire consumer journey. Now, we will be developing a full-stack web application using the MERN stack and Node.js, on which to implement this research work and deploy our trained models to make it available for the end-user.