DBTune - Serving music-related RDF since 2007

Linking Open Data Centre for Digital Music, Queen Mary, University of London

0. Disclaimer

This is a non-commercial site. The information contained in this site has been collected from several sources and its accuracy cannot be guaranteed. No copyright has been intentionally violated. If you feel a copyright has been violated please contact me immediately and the material will be removed

1. Introduction

DBTune hosts a number of servers, providing access to music-related structured data, in a Linked Data fashion. It now provides access to more than 14 billion RDF triples. All the services hosted here are based on open Web standards such as RDF and SPARQL.

This effort is part of the Linking Open Data on the Semantic Web community project, hosted by the Semantic Web Education and Outreach interest group.

You can browse DBTune data using any of the following Semantic Web browsers:

You can also use the Tabulator Firefox extension or the OpenLink Data Explorer.

2. Quick overview of underlying concepts (or the "Semantic Web" in 10 lines... Don't slap me, please)

Resources on the Web can be far more than web pages. They can identify anything: me, a French band, an audio signal, etc.

Such resources have associated representations (accessed, in our case, through HTTP), which may be either human-readable (an XHTML document, for example) or machine-processable (Microformats or RDF). These representations may hold links to further resources, allowing to jump from one resource to another, which may be actually hosted in different places. This is were the Web aspect comes into place.

Let's take an example. The resource http://dbtune.org/jamendo/artist/5 identifies a French band, called Both. Asking for a RDF representation of this resource gives us back the following statement:

<http://dbtune.org/jamendo/artist/5> <http://xmlns.com/foaf/0.1/based_near> <http://sws.geonames.org/2991627/>
We can then follow this link to get to a resource identifying a geographic location, located within the dataset. Moreover, this link is also a resource identifier! So if we don't know what it means to be based_near something, we can still access a representation of it, which would give us statements such as "based_near is a property, and it relates a person or a group of person to a geographical location".

3. Available datasets

The datasets available on this server, as well as their interlinks, correspond to the blue circles in the following diagram.

Depiction of the different DBTune datasets

Original diagram from Richard Cyganiak under the CC-BY-SA license.

The following datasets are available on this server:

  1. Jamendo, holding data from the Jamendo website (SPARQL end-point available, as well as browsable URIs, RDF dump available) ;
  2. Magnatune, holding data from the Magnatune label (SPARQL end-point and browsable URIs, RDF dump available) ;
  3. BBC John Peel sessions, holding data released during the Hackday, 2007 (SPARQL end-point and browsable URIs, RDF dump available) ;
  4. A chord symbol service, holding descriptions of chords, generated on-the-fly (browsable URIs)
  5. AudioScrobbler data, displaying your last 10 played tracks as RDF linked to Musicbrainz dereferencable URIs
  6. MySpace data, providing URIs and associated RDF representations for top-friends and available tracks on MySpace
  7. Musicbrainz data, powered by D2R server and a D2RQ mapping, SPARQL end-point available
  8. BBC playcount data, linking The BBC Programmes linked data and the Musicbrainz linked data, SPARQL end-point available
  9. Echonest Analyze XML to Music Ontology RDF transform
  10. Henry, a SPARQL end-point interpreting signal processing workflows and providing on-demand content-based data

4. Links to external datasets

The following dataset interlinking have been achieved:

  1. Jamendo to Musicbrainz (using some kind of record linkage technique)
  2. Jamendo to Geonames
  3. Magnatune to DBPedia
  4. John Peel to DBPedia
  5. Musicbrainz to DBPedia, MySpace and Lingvoj
  6. BBC playcount data to BBC Programmes and Musicbrainz

5. Acknowledgements

All the things provided by this server relies on the work of many people. Let's try to mention a few:

6. Contact

Yves Raimond, Centre for Digital Music, Queen Mary, University of London

E-mail: yves __at__ dbtune __dot__ org