Recommendation systems play a crucial role in personalizing user experience by suggesting products, services, or content based on user preferences and previous behavior. The traditional recommendation systems, while being effective, often lack interpretability and transparency, leaving the user without insights into how recommendations are derived. This paper proposes a user review-based explainable recommendation system that utilizes SHapley Additive exPlanations (SHAP) to enhance the interpretability of recommendations. By applying models like BERT-based collaborative filtering XGBoost and AdaBoost model with user reviews from the Yelp dataset, we generate personalized recommendations while offering feature-level explanations. The XGBoost model demonstrated superior performance, with a precision of 0.90 and NDGC of 0.94. SHAP visualizations such as summary, interaction, and dependency plots reveal the impact of features and its depending factors on the model prediction. The results indicate that the features impact the model through complex, nonlinear relationships and interactions, and the influence of one feature does affect and modify another feature’s influence. Integrating SHAP into recommendation systems improves both models’ accuracy and user trust and significantly enhances the efficiency and explainability of recommendation systems.