Research on Music Recommendation Model with Limited Historical Data and User’s Information

Nowadays, music streaming services are becoming more popular with the emergence of smartphones. For instance, about 2 million users used Spotify, which is one of the largest music streaming applications in the world, in 2010 and about 180 million users are on Spotify now. These music streaming applications also provide recommendation char to users for better satisfaction on application. For example, in YouTube music, there is a “Releasing Stress” chart. Also, there is a music list which people made and are able to share with other people who are in a similar situation. However, these recommendation lists are subjective, and this might cause side effects because some users do not choose their favorite artists or do not set their age when they use the streaming site first. Hence, in this paper, we are going to evaluate the music trend of these days and collect the various music streaming data from different users. Moreover, we will implement the pre-trained music recommendation model based on the limited users’ information such as their age, gender, religion, home country, and current location etc.