Enhancing Product Perception: A Novel Meta-Model Approach for Sentiment Analysis in Product-Based Reviews

Abstract

The proliferation of online purchasing and product evaluations, along with the rapidly expanding e-commerce industry, poses a significant problem for businesses looking to derive actionable data. High-star ratings are common, yet essential negative evaluations are sometimes overlooked. The urgent need for efficient sentiment analysis methods to enable businesses to extract sentiment from reviews and ratings is discussed in this study. By utilizing these approaches, companies may increase their comprehension of consumer input and make well-informed decisions that will lead to better product development and customer happiness. Sentiment analysis aids efforts to enhance products by revealing user perceptions of the items. However, traditional machine learning-based sentiment analysis techniques are unreliable and come with a hefty computational cost. While deep learning has demonstrated encouraging progress in sentiment analysis techniques, efficiently optimizing hyperparameters is still a hurdle. To determine which supervised machine learning classification technique produces the most reliable sentiment analysis results, this paper compares several supervised machine learning techniques used on online product reviews with the meta-model, which combines neural networks with support vector machines that have been proposed. With an amazing accuracy rate, the model reported in this study outperformed conventional methods. These findings highlight the usefulness of sentiment research in clarifying client attitudes and enhancing goods from a commercial perspective.