Automatic Playlist Recommender

CS 109 A Final Project

CS 109 A Final Project

Group 54: Yalda Amini, João Araujo, Daniel Barjum Dec 2018

Problem statement and motivation

Our problem question is as follows:

“Can we predict the class of a playlist, and use the features of its songs, to recommend appropriate new songs to that playlist?”

Music recommender system have become an important part of the music streaming industry recently, and it contributes to their revenure vastly, by increasing the user engagement and satisfaction from the service. One of the primary products of the Spotify, Pandora and Apple music is playlists, which is the users list of the his/her selected songs based on his/her mood. And this music recommender systems tries to add more relevant songs to the user’s current playlist based on the feature of the current songs or the user’s profile. During this project we don’t have access to user’s profile data, but we can use other datasets to do the classification of the palylist, in order to suggest most similar new songs to that playlist from the a pool of songs.