A HYBRID APPROACH TOWARDS A MOVIE RECOMMENDATION SYSTEM

Abstract

Online streaming media platforms are booming these days. OTT (Over-The-Top) platforms like Netflix, Amazon Prime, Hotstar, etc., provide online services that cater specifically to their users and provide them with some much-needed entertainment, which helps them to relax in this extremely busy world and spend some quality time with their friends and family. But when it’s time for a short break just to freshen up the mind, everyone prefers choosing from the options that are just a click away, what could be better than a user being recommended their favourite songs to play, movies to watch during a break. This is where the highly advanced and convenient recommendation systems come into play. Recommendation systems solve this problem by analyzing the large volume of dynamically generated data to provide personalized product suggestions and services to the users. With the day-to-day increasing popularity and demand of such services, and with increasing competition in this massive service-oriented sector, a highly functional and effective recommendation system could provide a significant marketing edge to a company over its competitors, as wide range of products, services and their substantial amount of information are available on the service provider company’s website and as a result, the users struggle to find relevant information matching their preferences. Thus, we have tried to design a framework for a movie recommendation system that will recommend movies to the user based on their preferences and user content history. First, we have taken a dataset provided by TDMB movies. Then, we have used and tested various classical machine learning models like Content-Based Filtering and Collaborative Filtering to recommend the best movies based on user ratings, user watch history, etc., that are more likely to be watched by the customer. We have also combined these models to create a Hybrid model which provides better recommendations. This could help in improving the revenue earned by the media platforms, which in turn is the main purpose of any recommendation system.