• Accidentally deleted an artist or track? Read this...

    1 Jun 2010, 06:52 by evilrix


    We'd like to announce a small tool we've written for the playground that allows you to restore any artist or track you may have deleted from your library by accident.

    It's not all bells and whistles, but it might be useful if you've ever removed something from your library only to have that gut wrenching feeling when you realize you've deleted more that you intended.

    Here is the blog post announcing it formally.

    Hopefully you'll find this useful but if you do find any issues of have any feedback please do let us know by posting on the blog.

    All the best.

  • New Toys

    13 Oct 2009, 16:39 by E1i45

    In praise of our loyal subscribers we've added a subscriber only VIP zone to playground with some new toys. We'll be adding more to our VIP area in the future. As usual we'd love to hear your feedback as it helps shape the future of

    The new demonstrations are:

    Image Charts
    Creates a collage of artist images where the size corresponds to how often you've been listening to the artist. You can adjust the time frame from when your scrobbles are counted and other settings.

    History Charts
    Graphs for your top artists showing how often and when you listened to them.

    Artist Connections
    Ever wondered how Paris Hilton is connected to Metallica? Find out how artists are connected through the similar artists network.
  • Some "Multi Tag Search" Stats

    23 Oct 2008, 08:30 by klbostee

    The "Multi Tag Search" has been on Playground since day one, but still lots of people seem to be using it. We keep receiving new queries every day, adding up to not much less than 20 thousand unique queries so far. Here are a few clever examples:

    * beautiful guitar instrumental
    * acoustic atmospheric melancholic
    * hilarious cover
    * "one hit wonder" 90s
    * american "guilty pleasure"

    A quick and dirty classification of the 250 most frequently occurring tags revealed that:

    * 51% are genres (e.g. , )
    * 26% are styles (e.g. )
    * 7% are locales (e.g. , )
    * 5% are instruments (e.g. , )
    * 4% are moods (e.g. , )
    * 3% are times (e.g. , )
    * 2% are opinions (e.g. , )

    The remaining 2% are tags like and . Together with the many e-mails and forum posts, this led to a way bigger pile of feedback than what we were hoping for. You guys really did an amazing job! We learned a lot from it, and we are definitely planning to take this knowledge into account.

    If you're interested in this stuff, you might want to have a look at this
    poster by the way. We used it as wall decoration during our demo about the "Multi Tag Search" at ISMIR, a conference where researchers who work with music-related data meet. Elias also teamed up with Paul from Sun Labs to give an awesome tutorial about social tags at this conference, and he participated in the interesting panel on "Commercial Music Discovery and Recommendation" as well.
  • User-to-user Recommendations Charts

    30 Jul 2008, 07:14 by klbostee

    It's time to announce the first Playground casualty. We actually liked this demo a lot, but you guys seemed to disagree since it has consistently been the least popular demo for a while now. However, we're really just making place for new demos to come by removing this one, so stay tuned!
  • Multi Tag Search iGoogle Gadget

    19 Jun 2008, 15:31 by klbostee

    Go get it!
  • Most Unwanted Scrobbles

    5 Jun 2008, 17:38 by klbostee

    Once again, we added some charts on playground. This time they show the tracks and artists that were most frequently deleted by the community from their scrobbles. As always, feel free to let us know what you think!
  • Top User-to-user Recommendations

    28 May 2008, 13:01 by klbostee

    We just launched a new playground demo. It's a weekly chart of tracks and artists recommended by users to users. Let us know if you like it!
  • Islands of Music

    20 May 2008, 14:26 by E1i45

    The islands of music playground demonstration is something like a tag cloud where similar tags are located close to each together. The map was created using clustering algorithms.

    These algorithms group similar music on islands. Similar islands are placed close to each other. For example, various flavours of metal are located close to each other in the upper right of the map. The map also suggests several more or less continuous transitions. For example, there is a path from folk to doom metal (via psychedelic, progressive rock, and progressive metal).

    Another somewhat curious example is the sea of mistagged artist where various flavours of non-English world music can be found. Generally, not all clusters make sense and part of the explanation is that there is plenty of noise in the data.

    In more technical terms:

    The map is a self-organizing map of 13,000 randomly sampled users labelled with tags and artists associated with each user.

    Each of these 13,000 users is described with a tag cloud which is extracted from the music the user listens to. This data is normalized in a similar way as described here . One consequence of this is that a large part of alternative indie rock pop is averaged out.

    After all the normalization and pre-processing the 13k users are represented by 2000 distinct tags resulting in 13k sparse vectors in a 2k-dimensional tag vector space.

    Using singular value decomposition (SVD) the dimensionality is reduced to 120 dimensions. This 120-dimensional space is a latent semantic space in which no distinction is made, for example, between brazil and brasil.

    Using k-means clustering 400 prototypical users are computed. Users very close to the zero-vector are not considered for further analysis. (Given the normalization and the latent space mapping, these zero-vector users can be interpreted as either very average users, or so unique that they can’t be described within the 120-dimensional space.)

    Using a self-organizing map (SOM) the latent space is mapped to a 2-dimensional visualization space. The SOM has a size of 20 rows and 40 columns. A smoothed data histogram of the SOM is computed and visualized so that clusters show up as islands.