Simplifying Digital Streaming: An Innovative Cross-Platform Streaming Solution
Debendra Muduli, Saroj Kumar Sahoo, Jayanta Kumar Mishra, Debasish Pradhan, Santosh Kumar Sharma · 2024
Our research aims to investigate how reinforcement learning methods can be applied to develop a movie recommendation system. We frame the issue of interactive recommendation as a contextual multi-armed bandit, where the system learns user preferences by suggesting new movies and receiving feedback. Through reinforcement learning, we address challenges such as balancing exploitation and exploration and tackling the coldstart problem. Additionally, we incorporate the novelty factor of movies into our model. We evaluate both content-based and collaborative filtering approaches which achieved accuracy 72.33(%), both of which demonstrate effective recommendation outcomes.