Movie Recommendation System using Hybrid Models, Agentic AI, and Prompt-based Approach

Current movie suggestion tools often fail to grasp the full context of what a viewer is seeking. Methods that rely on collective ratings or simple movie attributes offer limited personalization. This study introduces a new system that merges semantic analysis from advanced language models with a deep learning ranker to provide more adaptive and contextsensitive recommendations. The process starts by converting user queries and film descriptions into rich numerical representations. A primary ranking module then analyses these alongside user data to score potential matches. For new users, a separate mechanism promotes a varied selection across genres to gather initial preferences. A key feature is a natural language interface where users can express their mood, which the system translates into genres to guide its search. Tests confirm that this hybrid approach improves suggestion accuracy, diversity, and relevance over standard techniques, pointing toward a more intuitive and responsive recommendation experience.