HYBRID BOOK RECOMMENDER SYSTEM USING IMPLICIT FEEDBACK: A MACHINE LEARNING APPROACH

Abstract

A recommender system is a common instrument used by businesses to improve customer loyalty and revenues. The two most popular methods when implementing a recommendation system are collaborative filtering and content-based filtering, with the first offering recommendations based on user experience and the second utilizing characteristics of recommendable items.Since the efficiency of each recommendation model is constrained and each has its own strengths and disadvantages in the field of recommender systems, hybrid recommendation models are gaining more interest. The goal of the study was to propose a hybrid book recommendations system with implicit feedback and compares its ability to predict user ratings in an e-book application with basic recommender systems. The models were evaluated using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) measures. A hybrid model was built and compared to the pure models by integrating the two basic approaches.In addition, five models, two based on collaborative filtering, two based on content filtering and one hybrid were created. The results showed that by integrating both methods in a hybrid model with implicit feedback, the lower RMSE and MAE were achieved as compared to the collaborative filter model and content based model on RMSE and MAE measures.