Spotify Million Playlist Dataset and RecSys Challenge


Here at Spotify, we love playlists. Playlists like Today’s Top Hits and RapCaviar have millions of loyal followers, while Discover Weekly and Daily Mix are just a couple of our personalized playlists made especially to match your unique musical tastes.

Our users love playlists too. In fact, the Digital Music Alliance, in their 2018 Annual Music Report, state that 54% of consumers say that playlists are replacing albums in their listening habits.

But our users don’t love just listening to playlists, they also love creating them. To date, over 2 billion playlists have been created and shared by Spotify users. People create playlists for all sorts of reasons: some playlists group together music categorically (e.g. by genre, artist, year, or city), by mood, theme, or occasion (e.g. romantic, sad, holiday), or for a particular purpose (e.g. focus, workout). Some playlists are even made to land a dream job, or to send a message to someone special.

The other thing we love here at Spotify is playlist research. By learning from the playlists that people create, we can learn all sorts of things about the deep relationship between people and music. Why do certain songs go together? What is the difference between “Beach Vibes” and “Forest Vibes”? And what words (and emojis) do people use to describe which playlists?

The emoji most distinctively associated with the “most emoji-ed” artists from Spotify playlist titles, courtesy of Spotify Insights.

By learning more about nature of playlists, we may also be able to suggest other tracks that a listener would enjoy in the context of a given playlist. This can make playlist creation easier, and ultimately help people find more of the music they love.

Spotify’s “Recommended Songs” feature suggests songs to add to a playlist

To enable this type of research at scale, earlier this year we released The Million Playlist Dataset (MPD) to the academic research community. Sampled from the over 2 billion public playlists on Spotify, this dataset of 1 million playlists consist of over 2 million unique tracks by nearly 300,000 artists, and represents the largest dataset of music playlists in the world. The playlists were created by Spotify users between January 2010 and November 2017. Each playlist in the MPD contains a playlist title, the track list (including track IDs and metadata), and other metadata fields (last edit time, number of playlist edits, and more). All data is anonymized to protect user privacy. Here’s an example of a typical playlist entry:

{
        "name": "musical",
        "collaborative": "false",
        "pid": 5,
        "modified_at": 1493424000,
        "num_albums": 7,
        "num_tracks": 12,
        "num_followers": 1,
        "num_edits": 2,
        "duration_ms": 2657366,
        "num_artists": 6,
        "tracks": [
            {
                "pos": 0,
                "artist_name": "Degiheugi",
                "track_uri": "spotify:track:7vqa3sDmtEaVJ2gcvxtRID",
                "artist_uri": "spotify:artist:3V2paBXEoZIAhfZRJmo2jL",
                "track_name": "Finalement",
                "album_uri": "spotify:album:2KrRMJ9z7Xjoz1Az4O6UML",
                "duration_ms": 166264,
                "album_name": "Dancing Chords and Fireflies"
            },
            {
                "pos": 1,
                "artist_name": "Degiheugi",
                "track_uri": "spotify:track:23EOmJivOZ88WJPUbIPjh6",
                "artist_uri": "spotify:artist:3V2paBXEoZIAhfZRJmo2jL",
                "track_name": "Betty",
                "album_uri": "spotify:album:3lUSlvjUoHNA8IkNTqURqd",
                "duration_ms": 235534,
                "album_name": "Endless Smile"
            },
            {
                "pos": 2,
                "artist_name": "Degiheugi",
                "track_uri": "spotify:track:1vaffTCJxkyqeJY7zF9a55",
                "artist_uri": "spotify:artist:3V2paBXEoZIAhfZRJmo2jL",
                "track_name": "Some Beat in My Head",
                "album_uri": "spotify:album:2KrRMJ9z7Xjoz1Az4O6UML",
                "duration_ms": 268050,
                "album_name": "Dancing Chords and Fireflies"
            },
            // 8 tracks omitted
            {
                "pos": 11,
                "artist_name": "Mo' Horizons",
                "track_uri": "spotify:track:7iwx00eBzeSSSy6xfESyWN",
                "artist_uri": "spotify:artist:3tuX54dqgS8LsGUvNzgrpP",
                "track_name": "Fever 99u00b0",
                "album_uri": "spotify:album:2Fg1t2tyOSGWkVYHlFfXVf",
                "duration_ms": 364320,
                "album_name": "Come Touch The Sun"
            }
        ],

    }

Along with this dataset, we partnered with researchers from the Johannes-Kepler University Linz and the University of Massachusetts Amherst to launch the RecSys Challenge 2018, the annual data science challenge for the ACM Recommender Systems conference. The task for this year is automatic playlist continuation, where researchers are asked to submit their systems to correctly predict the songs we’ve hidden from real user-created playlists. The best submissions stand to win up to $4000 in two different competitive tracks. The competition is running now and will be open until the end of June 2018. If you are part of an academic research institution and are interested in participating, please visit https://recsys-challenge.spotify.com to sign up.

And if you love playlists too, and would like to work with us on solving problems like these beyond this challenge, we’re hiring!



Source link