Music recommender system using restricted Boltzmann machine with implicit feedback

Abstract

Now-a-days with explosive growth of the internet, the amount of information available overwhelms user. Collaborative filtering is the most popular method for solving this problem. Most of the research in recommender system concentrates on explicit ratings for their recommendation. The amount of implicit ratings available are much more compared to explicit ratings as implicit data are automatically generated when user interacts with the system. With the use of implicit feedback recommender system can compute preference of users without users giving rating to the system which is large is number. With the increase of computational power and popularity of deep learning algorithms, more and more research are carried out which are implementing deep learning method to provide recommendation. In this chapter we have shown a method which implements restricted Boltzmann machine for recommendation system. Here the number of times a user has listened to any music has been used as implicit feedback. We have also explained how to use contrastive divergence algorithm to train restricted Boltzmann machine and learn its parameters. Our model has shown better results by showing less error compared to existing collaborative filtering engine from Apache.