BBC Music talking semantics

Striking a chord with music fans

The BBC is working on an online project that uses open source and semantic web technology to create a comprehensive resource for music fans.

The BBC artists pages, a repository of information on singers and bands played on several BBC radio stations, have been running as a closed beta project since June last year.

But the project is more than just putting factual data onto a web page - it's about gathering information around an artist and using semantic web technology to link to other resources that give users more than just facts and figures.

Semantic web technology involves the tagging of basic factual information with metadata - adding context around the artist such as song names, previous bands, people they've collaborated with, venues they played at and so on - which then allows computers to establish relationships between the information.

Speaking to silicon.com, Matthew Shorter, interactive editor for music at the BBC, said: "This is part of a general movement that's going on at the BBC to move away from pages that are built in a variety of legacy content production systems to actually publishing data that we can use in a more dynamic way across the web."

The use of metadata makes the artists' pages more useful by providing access to as much relevant content as possible from a diverse range of sources.

Without linking and contextualising data in this way, it's much less likely people will find exactly what they want when trying to find out more about a band or artist.

For example, currently, if you look at the entry for Lily Allen and select the 'related artists' section it shows she is the daughter of actor Keith Allen. By clicking on his name, you can see that he is credited on records by New Order and Black Grape as well as access his Wikipedia entry.

Compared to a conventional music resource, the way in which the data is linked together takes you to other information that you might not know but which might be of interest.

"The principle behind it is, let's try and drill down to the stuff that people are interested in. We're really interested in - instead of just physical releases - building in a concept of an album [or artist] as a kind of cultural entity," Shorter said.

The BBC artists' pages draw their metadata from online open source 'metadatabase' MusicBrainz - which has information on more than 400,000 artists - to aggregate as much relevant information on a particular artist as possible.

Once the artist has a MusicBrainz ID, the contextual information around them is imported to their BBC page. "By assigning a MusicBrainz ID to an artist that also enables us to go off and fetch the Wikipedia biography and pull that into the page," Shorter explained.

According to Shorter, reusing data from MusicBrainz and Wikipedia also provides better value for the licence payer as the BBC isn't wasting resources reproducing data already in the public domain.

Currently the beta version of the project consists of an artists page, which shows how often the band or musician has been played on BBC radio in recent weeks, as well as an album review section.

From both of these points, users can access information - much of which comes from MusicBrainz - including other BBC programmes featuring the artist or track, album reviews, blog posts and official artist pages.

The next stage for the project is a public beta at some point in the near future but for now Shorter's team is focusing on building tools that will allow other BBC production teams to add data to the project.