Tоwards Орtimizing Hуbrid Mоvie Reсоmmender Sуstems

Abstract

Research in movie recommendation systems addresses several specific challenges to enhance the accuracy, relevance, and user satisfaction of the recommendations. Some of the key challenges include sparsity of data and cold start for new movies. Integrating collaborative filtering and content-based filtering in a seamless and effective manner also poses another key challenge. This article explores strategies for optimizing hybrid recommendation systems to leverage the strengths of each approach. By addressing these three key challenges, we aim to advance the state of movie recommendation systems, providing more effective and user-centric solutions that cater to the diverse preferences and needs of movie enthusiasts. Similarity-based models (memory-based) combined with matrix factorization-based models (model-based) were used to generate an optimal hybrid movie recommendation model. We evaluated the models using a large dataset of movies and we showed that the proposed model is both efficient and scalable.