Analyzing Amazon Customer Sentiments using Recurrent Neural Network Architecture

Abstract

With the exponential growth of e-commerce, understanding customer sentiments has become increasingly crucial for businesses. As one of the largest online retailers, Amazon generates a vast volume of customer reviews daily, offering valuable insights for companies to enhance their products and services. In this research, a system for analyzing customer sentiments on the Amazon platform is proposed to manage valuable insights into customer preferences and opinions. The main goal of this paper was to find effective marketing strategies to drive profit growth, monitor market trends, and perform competitive analysis. The achieved results by the proposed system were accurate in predicting the sentiments of Amazon customers. The size of the testing set was 1200 for positive reviews and 1200 for negative reviews. The system predicted 995 true positives and 205 false positives for positive reviews, while for negative reviews, it predicted 861 true negatives and 339 false negatives.