Cross-Content Recommendation between Movie and Book using Machine Learning

Abstract

Machine learning-driven recommendation systems are widely used in today’s growing digital world. Existing movie and book recommender systems work using a collaborative approach, which can result in a lack of fresh and diverse content and a reduced surprise factor. There is also no platform providing recommendations across different contents, such as recommendations for books from movies and vice versa. In this paper, our main goal is to introduce a cross-content recommendation system based on the descriptions of movies and books and identifying similarities using natural language processing and machine learning algorithms. We processed a combined dataset of the two different types of contents, generated a TF-IDF vector of the descriptions and apply three different algorithms: K-means clustering, hierarchical clustering, and cosine similarity. There being no known cross-content recommendation research and no similar dataset with ground truth labels, we applied subjective reasoning to evaluate the results of our system.