Modelling a movie recommendation system using a mixed classification method

The primary objective of this research is to enhance the accuracy and reliability of movie recommendation systems (MRS) by addressing common challenges such as the cold start problem, data sparsity, and malicious attacks. This study introduces a mixed machine learning approach, combining traditional classification techniques with advanced models like autoencoders and Restricted Boltzmann Machines (RBMs). By incorporating trust metrics into the recommendation process, the system aims to provide more personalized and dependable movie suggestions. The methodology involves using a hybrid classifier that integrates Naïve Bayes, Support Vector Machines, autoencoders, and RBMs to improve recommendation precision. The results demonstrate that the inclusion of trust-based algorithms significantly enhances the recommendation accuracy, especially in scenarios with limited data. The proposed model outperformed traditional methods, achieving higher validation accuracy and effectively addressing the cold start problem. The integration of trust metrics with RBM further improved the system's performance, achieving an accuracy of 93.4%. This study concludes that hybrid machine learning approaches, combined with trust-aware mechanisms, offer a robust solution to the challenges faced by conventional MRS, providing a foundation for future research in developing more efficient and reliable recommendation systems.