<![CDATA[The rapid growth of the digital film industry has resulted in a significant increase in available content, causing users to experience difficulties in finding movies that match their personal preferences. This condition highlights the need for an effective and personalized recommendation system. This study proposes a movie recommendation system based on a Content-Based Filtering (CBF) approach using the Naive Bayes algorithm to generate recommendations according to movie content characteristics and users’ preference histories. The MovieLens dataset obtained from Kaggle is used in this research. The research process includes data preprocessing, user profile construction, implementation of the Naive Bayes-based CBF model, and performance evaluation using Mean Absolute Error (MAE), Precision@5, Recall@5, and Mean Average Precision (MAP). The evaluation was conducted through ten repeated experiments with different training and testing data splits to ensure the reliability and stability of the results. The experimental results show that the Naive Bayes algorithm is able to effectively model the relationship between movie content attributes and user preferences. The evaluation achieved an average MAE of 0.368, Precision@5 of 0.82, Recall@5 of 0.11, and MAP of 0.42. The relatively high Precision@5 and MAP values indicate that the recommended movies are highly relevant and well-ranked at the top positions, while the low MAE value reflects satisfactory rating prediction accuracy. These results demonstrate that the proposed Naive Bayes-based Content-Based Filtering approach is capable of producing relevant, personalized, and stable movie recommendations, particularly in scenarios involving new or sparsely rated items.]]>
Copyrights © 2026
