Music streaming platforms enable people to access millions of tracks using computers and mobile devices. However, users cannot browse manually millions of tracks to find music they like. Building recommender systems suggesting music fitting the current context of a user is a challenging task. A deeper understanding for the characteristics of user-curated playlists naturally contributes to more personalized recommendations. To get a deeper understanding of how users organize music nowadays, we analyze user-curated playlists from the music streaming platform Spotify. Based on the audio features of the tracks, we find an explanation of differences in the playlists using a PCA and are able to group playlists using spectral clustering. Our findings about playlist characteristics can be exploited in a SVD-based music recommender system and our proposed clustering approach for finding groups of similar playlists is easy to integrate into a recommender system using pre-or post-filtering techniques.
%0 Journal Article
%1 Pichl2017
%A Pichl, Martin
%A Zangerle, Eva
%A Specht, Günther
%C USA
%D 2017
%I IGI Global
%J Int. J. Multimed. Data Eng. Manag.
%K Analysis Learning Content Music Retrieval Data Information from:kamber Machine User-Generated
%N 4
%P 44–59
%R 10.4018/IJMDEM.2017100103
%T Understanding User-Curated Playlists on Spotify: A Machine Learning Approach
%U https://doi.org/10.4018/IJMDEM.2017100103
%V 8
%X Music streaming platforms enable people to access millions of tracks using computers and mobile devices. However, users cannot browse manually millions of tracks to find music they like. Building recommender systems suggesting music fitting the current context of a user is a challenging task. A deeper understanding for the characteristics of user-curated playlists naturally contributes to more personalized recommendations. To get a deeper understanding of how users organize music nowadays, we analyze user-curated playlists from the music streaming platform Spotify. Based on the audio features of the tracks, we find an explanation of differences in the playlists using a PCA and are able to group playlists using spectral clustering. Our findings about playlist characteristics can be exploited in a SVD-based music recommender system and our proposed clustering approach for finding groups of similar playlists is easy to integrate into a recommender system using pre-or post-filtering techniques.
@article{Pichl2017,
abstract = {Music streaming platforms enable people to access millions of tracks using computers and mobile devices. However, users cannot browse manually millions of tracks to find music they like. Building recommender systems suggesting music fitting the current context of a user is a challenging task. A deeper understanding for the characteristics of user-curated playlists naturally contributes to more personalized recommendations. To get a deeper understanding of how users organize music nowadays, we analyze user-curated playlists from the music streaming platform Spotify. Based on the audio features of the tracks, we find an explanation of differences in the playlists using a PCA and are able to group playlists using spectral clustering. Our findings about playlist characteristics can be exploited in a SVD-based music recommender system and our proposed clustering approach for finding groups of similar playlists is easy to integrate into a recommender system using pre-or post-filtering techniques.},
added-at = {2024-03-25T16:44:47.000+0100},
address = {USA},
author = {Pichl, Martin and Zangerle, Eva and Specht, Günther},
biburl = {https://www.bibsonomy.org/bibtex/2e2d152dadd349b9794eaa36068811fe6/sop2-ffzg},
day = 1,
doi = {10.4018/IJMDEM.2017100103},
interhash = {e9ded4d3c4191f4caef317fd8e0d3d68},
intrahash = {e2d152dadd349b9794eaa36068811fe6},
issn = {1947-8534},
issue_date = {October 2017},
journal = {Int. J. Multimed. Data Eng. Manag.},
keywords = {Analysis Learning Content Music Retrieval Data Information from:kamber Machine User-Generated},
month = {10},
number = 4,
numpages = {16},
pages = {44–59},
publisher = {IGI Global},
timestamp = {2024-03-25T16:44:47.000+0100},
title = {Understanding User-Curated Playlists on Spotify: A Machine Learning Approach},
url = {https://doi.org/10.4018/IJMDEM.2017100103},
volume = 8,
year = 2017
}