A Movie and Book Recommender System using Surprise Recommendation

Abstract

With viable amounts of global Internet users and their online shopping spree, the world has led towards an astonishing and exponential digital data growth. It is of at most importance that this data has to be properly consumed and utilized when it comes to providing personalized content to both users and service providers. This is where the role of a Recommender System comes into active participation. These systems being user-centric are used to provide information that suits to the needs and interest of a user. In this paper, two types of Recommender systems are proposed. The first one is a Movie recommender and the second a Book recommender. For the movie recommender, the MovieLens dataset is used and the personalized Book content is obtained applying various prediction algorithms available in Surprise Recommendation kit. We make use of the Books-crossing dataset that has the book relevant information along with its ratings. We use the collaborative and content-based filtering techniques for obtaining recommendations and the systems are also evaluated using different metrics like RMSE, test time and the error factor.

Keywords: MovieLens, Collaborative filtering, Content-based filtering, Books-crossing, Scikit, Surprise, Jaccard coefficient, Cosine coefficient, Test Time, Error, Prediction