User Song Preferences using Artificial Intelligence

With the rapid development of mobile devices and more people having internet access, people have access to music collections at an unprecedented scale. Music libraries can easily have more than 15 million songs, people will feel overwhelmed choosing from the ocean of song available. Thus, an efficient song recommender system is necessary for the music service providers and customers. The music streaming companies can attract and retain users with a good recommender system. In the music recommender system field, many music streaming companies are working on building high-precision music recommender system. Thus, this field have a high market demand for good quality music recommendation system. In this research paper, it is the initial stage of the research project using preference learning for the music recommendation system as it has potential to be utilized as optimum user song preferences for the music recommendation systems. This research project conducted preliminary experiment for training data set for few classification models that are Random Forest, Logistic Regression, Support Vector Machine, Naive Bayes, Decision Tree and K-Neighbors for that can be used for the collaborative model. From the accuracy output, it was observed that Logistic Regression gave the highest score that is 18.03% and the Random Forest gave lowest score that is 8.19%.