User Centric and Collaborative Movie Recommendation System Under Customized Platforms

Abstract

Topping the entertainment industry, movies have become a really life-changing factor for this industry. The only difficulty individuals are facing right now is to find the right content which fits their choice of demands. To solve this problem, recommendation systems have been brought into use. Various filters can be applied to find the content of your desire. The aim of this paper is to predict user’s movie rating along with genre in order to recommend movies. The movies are recommended based on user activity and content to which they potentially give high ratings. User choices and history makes it easier to predict movies and filter out content. Many methods have been used to implement a recommendation system, but in this paper, we have majorly used collaborative filtering for recommending and content-based filtering for sentiment analysis. It will be used to output a list of recommended movies based on an individual’s choice, ratings, likes and past behavioral pattern. In this paper, to improve accuracy and effectiveness, similarity score is used. It is a numerical value that ranges between 0 to 1 and helps to determine how much a group of items are similar to each other. Cosine similarity makes the process effective by quantifying the similarity in the form of angular distance. As compared to Euclidean distance, cosine similarity is used because the precision of the cosine angle and the equi-distance of movies are almost identical.