Enhancing Decision-Making in E-Commerce using Large Language Models : A Comparative Analysis of GPT-4 and LLaMa-3 for Aspect-Based Sentiment Analysis

The study investigates the effectiveness of Large Language Models, specificallyGPT-4 and LLaMA-3, in enhancing decision-making processes within the e-commerce sector by applying aspect based sentiment analysis. With the increasing importance of analyzing customer sentiment to improve customer satisfaction and engagement, the study focuses on how these models can support Decision Support Systems by automating the extraction and interpretation of sentiment data from customer content. A custom-labeled dataset was created and used to compare the accuracy, efficiency, and deployment costs of both models. The findings underscore GPT-4’s superior accuracy in sentiment detection across various aspects, although LLaMA-3 demonstrates notable efficiency benefits and cost effectiveness, especially in self-managed deployments. This research contributes to practical applications in e-commerce by offering insights into the integration of LLMs into DSS for improved customer insights and data-driven decision-making, as well as broader implications for AI-driven sentiment analysis in customer engagement strategies.