APML for Pandora
I've
been experimenting with creating APML (the Attention Profile Markup
Language) based upon available taste data. Last week I created a last.fm APML web service as well as a del.icio.us web service.
This week, I've created a web service that will serve up an APML taste
profile based upon your Pandora favorite artists. For example, the
URL:
http://research.sun.com:8080/AttentionProfile/apml/pandora/paul.lamere
will serve up the APML based upon my Pandora listening, while the URL:
http://research.sun.com:8080/AttentionProfile/apml/pandora/tconrad
will serve up Tom Conrad's (the CTO of Pandora) APML data.
This web service is fairly easy to create. Pandora creates an RSS feed for each user that shows the user's favorite artists. The APML web service grabs that RSS feed and adds each of the artists to the ExplicitConcepts section of the APML file. The next step is a bit trickier. We want to be able to turn this set of artists into a more abstract representation of a person's musical taste - to include in the ImplicitConcepts section. To do this, I rely on the social tags applied by last.fm users. For each artist in a user's set of favorite Pandora artists, the APML service collects the top 50 social tags that have been applied to that artist from last.fm. The tags are aggregated and weighted based on the user's affinity for the associated artist. I then take top scoring tags to generate the ImplicitConcept section of the APML.
Meager information on these experimental web services can be found on the taste broker page.