Consumer reviews significantly influence purchasing decisions on e-commerce platforms and social media. Many e-commerce sites utilize recommendation systems to address information overload, which helps users decide on products. Integrating user opinions into these systems is a burgeoning research area, highlighting the importance of sentiment analysis and opinion classification. This study explores enhancing recommendation systems by analyzing product review sentiments to tailor suggestions to consumer profiles. Using natural language processing (NLP), we examine the sentiment of review texts to determine their positivity or negativity. Logistic regression and Naïve Bayes algorithms assign scores to these sentiments. These polarity scores feed into a collaborative item–item filtering system during the recommendation phase, resulting in an innovative and efficient recommendation system. Our study employed Naïve Bayes, logistic regression, and K-nearest neighbors (KNN) algorithms. Our approach, tested on the Amazon database, demonstrated high-quality recommendations with an impressive accuracy rate of 94% and an error rate of 6%. Our findings revealed that logistic regression provided better precision than Naïve Bayes when the dataset was used for training and testing. This underscores the potential of combining sentiment analysis with recommendation systems to enhance consumer e-commerce experiences.