´óÏó´«Ã½

By now you may well have found the new ´óÏó´«Ã½ Music beta site - Matthew Shorter and have both blogged about it, and it's shown up on . If you haven't seen it yet, I strongly urge you read the blogs and take a look - it really is a huge step forward for ´óÏó´«Ã½ Music online, and for the data infrastructure of bbc.co.uk as a whole.

I'm not going to repeat Matthew and Tom here. Instead I'm going to explain how some of the thinking and technical features have come together in one of the really cool things we've been able to achieve - the data graphs on the new Artists Gateway and on Artist pages.

VCS, JSON and The Star
A large proportion of music played on ´óÏó´«Ã½ radio is played out through a system called VCS - basically an enterprise-scale mp3 player. As well as injecting music into the broadcast chain, it publishes data about the track now playing internally, and since September 2007 this internal feed has been archived.

The Music Discovery team have hooked the feed and archive, associated the tracks with schedule data and with MusicBrainz artist IDs, and injected the resulting data into /programmes. The immediate result is tracklists for each programme on programme pages.

We've had to roll tracklists back for the moment. They'll return in the near future.

But as well as individual programme views, /programmes can provide artist playout data - look at /programmes/music/artists/2fddb92d-24b2-46a5-bf28-3aed46f4684c.json for data on The Ting Tings and compare it with their beta page on /music. Yes the feed is public - you can hack with it too!

Back in March Michael blogged about five linked data sources at the ´óÏó´«Ã½.

Until this week /programmes was still the only node on the graph. This makes /music beta node number 2.

Data formats
Linked data means more than purpose-built data feeds between ´óÏó´«Ã½ systems, and more than making those feeds public. HTML is pretty good for making human-readable pages, but we all know it sucks when you start trying to screen-scrape data. So to help you do whatever you want with our data, we've made all the pages in the /music beta available in XML, YAML, JSON and RDF versions, and RDF versions of /programmes will be available soon.

Nick Humfrey and Patrick Sinclair are the semantic experts, and they'll be blogging about this in detail. But for more of the thinking behind it take a look at Nick's and Tom's presentation at Xtech.

Beta to live
At the moment links on pages only run from /music to /programmes, but not back the other way. That's because the existing Artist pages are still running alongside the beta. As soon as we come out of beta, the tracklistings on programme pages will link back to Artist pages on /music.

The main criterion for coming out of beta is the inclusion of tracklisting data for music played off CD or even vinyl but not out of VCS, which includes specialist shows and all of Radio 3. And that's what we're working on right now.