Author : Chetankumar Patel

Due to the fast-growing technological age, Artificial Intelligence (AI) has become a revolution in the electronic business (e-commerce) sector. The AI-based engine also allows e-commerce websites to provide customers with greater customer experience by using intelligent search capabilities, a personalized product suggestion system, inventory management, and automated customer service support systems like chatbots. This paper shows a critical comparative study of classical and advanced learning models, such as Support Vector Machine (SVM), Logistic Regression (LR), BERT, Random Forest (RF) and the proposed XGBoost and Gated Recurrent Unit (GRU) models, to enhance sentiment-driven recommendation systems on the Amazon Reviews dataset. The data is processed extensively in preprocessing, feature extraction and class balancing so that the model is trained well. The experimental performance of the traditional models like SVM and LR suggests a moderate level of performance, but the advanced models like BERT and RF have the higher level of accuracy. The proposed XGBoost model has an accuracy of 92.6% and the proposed GRU model has the highest accuracy of 95.87, which is essentially able to use sequential and contextual information in customer reviews. These results confirm that deep learning-based models especially GRU are very strong to perform better in recommendation systems and also help in creating a more tailored and fulfilling online customer experience.

Keywords: Artificial Intelligence, Customer Behavior, Customer Satisfaction, E-Commerce, Machine Learning.