Comparison between content-based and collaborative filtering recommendation system for movie suggestions

Abstract

In this current and recent decade, various data and information are actively collected. The problem with increasing data from year to year has made it difficult for people to make the right decision. This is because when data increased, the number of options to choose from will also increase. Therefore, recommendation systems are needed to address this problem and help recommend to users some options that meet their desirable requirements only. In this study, recommendation systems in the field of filming were conducted to provide movie recommendations services for users by using the content-based recommendation system and collaborative filtering recommendation system. For content-based recommendation system, movie recommendations are done by looking for similarities between active user profiles and movie genres. The similarities between active user profiles and movie genres are calculated by using the cosine similarity measure. For collaborative filtering recommendation system, movie recommendations are made by calculating the predicted rating for active users based on the rating values from their nearest neighbours. The nearest neighbours are identified by calculating the cosine similarity measure between the active users and existing users in the data set. Comparisons for these two recommendation systems are performed to identify which system works best in recommending movies to active users. The results show that the collaborative filtering recommendation system is more suitable in recommending movies to active users because this system is more successful in producing desirable recommendations compared to content-based recommendation system.