Optimizing Hybrid Book Recommendation Systems: A Comparative Analysis of Learning Rates and Embedding Dimensions

In this study, a hybrid recommendation system is analyzed by integrating collaborative filtering, content-based filtering, and deep learning techniques. The system considers user-item interactions and content features to enhance the accuracy and relevance of book recommendations. A comprehensive dataset was preprocessed, and mock user ratings were generated to simulate real-world scenarios. The model was designed using embedding layers, dense layers, and dropout regularization, with performance optimized through hyperparameter tuning. Evaluation metrics such as precision, recall, F1-score, root mean square error (RMSE), mean absolute error (MAE), coverage, and novelty were used to assess the model. Results demonstrated that a learning rate of 0.0005 with an embedding size of 50 provided the best balance of performance and efficiency. This hybrid approach significantly outperformed traditional methods, showcasing its potential for improving recommendation quality in practical applications.