Abstract—Recommender systems are algorithms designed to provide personalized suggestions for products and services that are most relevant to individual users. These systems analyze historical data about user interactions, such as reviews, clicks, likes, and dislikes, to predict and recommend products or services that align with users’ interests. By leveraging this data, recommender systems can effectively tailor suggestions to each user, enhancing their overall experience and satisfaction. Recently, sentiment analysis has gained traction as an approach for recommendation systems, offering an alternative to traditional methods such as collaborative filtering, content-based approaches, and hybrid models. By analyzing user sentiments expressed in product reviews, these systems can make recommendations based on the emotional tone of the feedback, leading to more personalized and contextually relevant suggestions. So, to enhance the performance of recommendation systems and address the cold start problem for new users and data scalability issues, we propose a hybrid approach that integrates sentiment analysis with collaborative filtering. This system leverages deep learning algorithms alongside traditional collaborative filtering techniques, such as Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN). We analyzed user reviews and comments using IMDB, Amazon, and Amazon Fine Foods datasets for experiments and results. Our approach recommends the best items and improves overall system accuracy, resolving cold start problems and data scalability. The results demonstrate that integrating sentiment analysis significantly enhances recommendation quality compared to traditional collaborative filtering methods and effectively mitigates the cold start issue.
 
Keywords—sentiment analysis, deep learning, recommender system, opinion mining, Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), collaborative filtering, factorization matrix, users’ ratings, cold start


Copyright © 2025 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).