2018/07/12 by Andreu Vall, Massimo Quadrana, Vall, Andreu +5
Computer Science · #Music and Audio Processing #Music Technology and Sound Studies
paper · pdf · doi:10.48550/arxiv.1807.04690
The automated generation of music playlists can be naturally regarded as a\nsequential task, where a recommender system suggests a stream of songs that\nconstitute a listening session. In order to predict the next song in a\nplaylist, some of the playlist models proposed so far consider the current and\nprevious songs in the playlist (i.e., the song context) and possibly the order\nof the songs in the playlist. We investigate the impact of the song context and\nthe song order on next-song recommendations by conducting dedicated off-line\nexperiments on two datasets of hand-curated music playlists. Firstly, we\ncompare three playlist models, each able to consider a different song context\nlength: a popularity-based model, a song-based Collaborative Filtering (CF)\nmodel and a Recurrent-Neural-Network-based model (RNN). We also consider a\nmodel that predicts next songs at random as a reference. Secondly, we challenge\nthe RNN model (the only model from the first experiment able to consider the\nsong order) by manipulating the order of songs within playlists. Our results\nindicate that the song context has a positive impact on the quality of\nnext-song recommendations, even though this effect can be masked by the bias\ntowards very popular songs. Furthermore, in our experiments the song order does\nnot appear as a crucial variable to predict better next-song recommendations.\n