Preference-Aware Music Recommendation Using Song Lyrics

Music is closely related to our lives. People enjoy music every day and share or express their emotions through music. Therefore, finding and recommending a favorite song among many kinds of songs is important and difficult at the same time. As the online music platform became popular and many people began to use music recommendations. Many approaches have been proposed to understand the user’s music preferences and various factors have been used. Also, various recommendation methodologies have been used, such as content-based (CB) or collaborative filtering (CF). In this work, we propose a recommender system using user’s preference that was defined based on the feature extracted from the context information of the song’s lyrics. We extract and use the latent-topic in the context of the song lyrics. Our results show that it is efficient to recommend songs based on the preferred singers by analyzing user preferences.