Enhancing Customer Experience through E-commerce Review Analysis: Using Topic Modeling and Rule Induction for Understanding User Perception

Abstract

The rapid growth of e-commerce has significantly altered the dynamics of business-consumer interactions, with online reviews becoming a crucial determinant of customer satisfaction. This research endeavors to delve into the realm of e-commerce reviews by employing advanced text analytics techniques such as Latent Dirichlet Allocation, rule-based induction, and sentiment analysis using Python programming language. The accuracy of the analytical model is rigorously evaluated through multiple linear regression analysis, root mean square error, and R-squared methods. To acquire comprehensive insights, the study collected textual review data from prominent e-commerce platforms like Amazon, unraveling valuable information regarding customer sentiments, behaviors, and the underlying factors that influence user perceptions. Through meticulous analysis, the research attains an impressive R-squared value of 0.7553, signifying the model's robustness in capturing the variance in the dataset. Furthermore, the root mean square error of 0.000096129 (9.61296520076313e-05) underscores the efficiency of the model by demonstrating a minimal margin of error. The findings of this study not only contribute significantly to the academic discourse surrounding e-commerce analytics but also bear practical implications for the industry. By uncovering critical factors that shape user perceptions and decision-making processes, businesses can implement strategic measures to enhance customer experience and drive profitability. This research serves as a valuable resource for e-commerce entities seeking to optimize their operations, refine marketing strategies, and foster long-term customer satisfaction.