Spotify Million Playlist Dataset

dataset·playlist recommendation·active·also: MPD

One million anonymized public playlists released for automatic playlist continuation research.

Verified facts

Official site	https://research.atspotify.com/2020/09/the-million-playlist-dataset-remastered ↗
Geography	United States
size	artists: nearly 300000 · playlists: 1000000 · unique tracks: over 2000000 · source public playlists: over 2000000000
scope	Music recommendation and automatic playlist continuation.
genres	user-created mixed playlists
license	Non-commercial open research terms
steward	Spotify
creators	Ching-Wei Chen; Paul Lamere; Markus Schedl; Hamed Zamani
languages	multilingual playlist titles
modalities	JSON playlists; playlist titles; track and artist identifiers; playlist metadata
geographies	Global
access method	AICrowd open-ended challenge after 2020 rerelease.
intended uses	automatic playlist continuation; music recommendation
related papers	RecSys Challenge 2018: Automatic Music Playlist Continuation
prohibited uses	Commercial use is not allowed under the published research terms.
annotation method	Playlist composition and metadata preserved; challenge hides tracks for continuation evaluation.
collection method	Sampled from public Spotify playlists created between 2010 and November 2017.
provenance claims	Spotify states data were anonymized to protect user privacy.
current availability	Research challenge page exists; access requires terms and may change.
commercial use limits	Commercial use prohibited by challenge terms.
us market scope basis	Publicly available or materially used in US-facing music/audio-AI research.
limitations biases disputes	Historical Spotify users and public playlists are not representative of all listening.; Playlist titles and track combinations can carry privacy or sensitive-context signals.