Deep Collaborative Movie Recommendation Systems

Abstract

Collaborative filtering driven recommendation systems have become significant in newlinediverse domains for their capacity to provide personalised recommendations. In e- newlinecommerce, these systems analyse users browsing histories and purchasing behaviours newlineto suggest relevant items. In the entertainment industry, collaborative filtering plays a crucial part in assisting platforms like Netflix and Spotify in suggesting movies, shows,and songs by analyzing users historical preferences and ratings. This technology also newlinefinds significance in online education, where it helps to suggest relevant courses and newlinelearning materials based on user interests and previous learning behaviour. Although newlinemuch research has been done in this domain, the challenges of sparsity and scalabil- newlineity in collaborative filtering still exist. Data sparsity refers to too few preferences of users on items, and hence it would be difficult to understand users preferences. Recommendation systems must keep users engaged with fast responses, and hence there is a challenge in handling large data, as these days it is growing quickly. Sparsity affects recommendation accuracy, while scalability influences the complexity of processing the recommendations. Matrix factorization, a traditional collaborative filtering algorithm,addresses the challenge of sparsity. However, matrix factorization does not fully capture user-movie interactions because it uses a simple dot product while predicting the newlinerating. Most of the research nowadays deep learning to exploit the nontrivial and non- newlinelinear user-movie relationships. Even though deep learning techniques are non-linear, newlinethey suffer from high variance and are prone to overfitting, which can affect the model s newlinegeneralisation capacity. newlineThe motivation behind this thesis is to design efficient algorithms to address sparsity and scalability problems, which, in turn, provide a better user experience and increase user satisfaction. Initially, a novel deep matrix factorization-based collaborative filtering algorithm calle