PRS-UBR: Product Recommender System Using Utility-Based Recommendation

E-commerce has become essential in today’s world as it provides global access, unprecedented convenience, and accessibility to both consumers and businesses. It offers a shopping experience that is available 24/7, without any physical store limitations, allowing customers to easily browse and purchase products from the comfort of their own homes. Recommender systems are crucial for enhancing user experience and minimizing information overload. It is achieved by utilizing user preferences and sentiments, thereby increasing the success of e-commerce platforms in a highly competitive and rapidly changing online market. The proposed methodology includes sentiment analysis and the development of a recommendation model. The sentiment analysis dataset is optimized through data preprocessing steps and for sentiment analysis, three machine learning models namely Random Forest, Logistic Regression, and XGBoost are used; the best-performing model is chosen for the recommendation system. User-based and item-based collaborative filtering models for recommendation systems are created, using adjusted cosine and cosine similarity metrics. The methodology concludes with an assessment and choice of the best recommendation model, providing a thorough method for sentiment analysis and customized product recommendations in the e-commerce space.