Music Recommendation System
Published:
Abstract
Music is an important component of the development of human culture and often holds itself as an identifying or character-building factor for many individuals. With the advent of technology, recommendation systems for music have become commonplace for the average listener. However, unfortunately, many of these systems or platforms often cannot fulfill the emotional requirements of individuals; instead often opting more for filtering or history-based algorithms in comparison to user mood-based recommendation systems. In this paper, we propose a modified mood-adapted content-based filtering model to provide a recommendation system using the Spotify API based on a decision-tree mood-labeled song dataset along with a dataset of users and details about the songs that they have listened to. We were able to develop a mood-adapted content-based filtering model that was successfully able to match the accuracy of more traditional models, thus able to show that we were able to incorporate adaptive factors into content-based filtering models without a loss in performance.