A Personalized AI Assistant for Analyzing and Authoring Product Reviews using Groq/Llama3-70B

This paper presents a novel AI assistant designed to streamline the analysis and creation of product reviews by leveraging modern web extraction, natural language processing, semantic vector indexing, and a state-of-the-art large language model via the Groq API for Llama3-70B. Our approach gathers review data from various online marketplaces, cleans and indexes the information, and then provides users with interactive query responses as well as personalized review drafting. A central con-tribution is our fine-grained aspect-based sentiment evaluation combined with a hybrid Retrieval-Augmented Generation (RAG) framework that ensures the generated outputs are both relevant and factually grounded. Experimental evaluations indicate that the system achieves a sentiment classification accuracy of 94.3 % while reducing the decision-making time by 67 % relative to conventional methods.