A novel recommendation framework using sentiment and rating weighted association score framework
Chaitali Choudhary, May El Barachi, Manoj Kumar · International Journal of Computers and Applications · 2026
Most current techniques focus on either ratings or reviews, with an emphasis on reviews. In this paper, we introduced an algorithm named SARWAS which uses ratings and reviews in the recommendation system. Deep learning model discussed in this work uses sentiment score and weighted rating association score (SARWAS) framework to combine these two factors. We gathered user evaluations from amazon web page and assessed the sentiment of each review. Subsequently, we used deep learning model to first assign random weights, followed by the computation of optimal weights via model training, ultimately producing a composite score for each product. We assessed the model by analyzing the association among ratings, reviews, and suggestions, as well as other metrics. Our results demonstrate that the suggested technique attains high accuracy. The connection between reviews and suggestions exceeds that between ratings and recommendations, with accuracy and precision at 95% and 89%, respectively.