raimond.me.uk
Yves Raimond's home
HomeDBTune

This page is outdated. Relevant information about the DBtune project is available at http://dbtune.org/.

Introduction

This effort has started in the context of the Linking open data community project of the Semantic Web Education and Outreach Interest Group.

Its main purpose is to make available freely available data concerning music on the semantic-web, such as Magnatune, Jamendo, Dogmazic, Mutopia, and to create links between them and other available semantic web repositories, such as Frederick Giasson's Musicbrainz dump, or DBPedia and GeoNames.

Motivation

We believe that by providing a large amount of inter-linked RDF information, thus accessible by tools such as the SWI semantic-web client (shameless self-promotion:-) ), the semantic-web client library (JAVA), or browsers such as the Tabulator or DISCO, we can demo a great potential of semantic-web technologies.

We can indeed imagine nice applications, such as a Geo music player, where you click on a world map, and the applications get information from the semantic-web in order to generate you a playlist from this region. We can obviously imagine the same sort of semantic media player for other criteria such as composition time, performance time, automatically extracted features (key, tempo, harmony...). Such applications are described in this blog post.

An other purpose is to create a large amount of information available for re-use and processing by the Knowledge Machine framework. Indeed, we could imagine that a semantic media player such as the one described before could in fact grab dynamically generated features, coming from a Knowledge Machine somewhere! And that you could visualise annotations and features, using the Sonic visualiser!

Technologies

In order to make this data available, we use D2R Server, along with P2R and URISpace.

We use the Music Ontology in order to describe music-related data.

So far...

The magnatune dump has been made available. This relies on the magnatune SQL dump, and uses this D2RQ mapping. There are not much information in the magnatune SQL dump, so we couldn't get behind the description of the Performance (ie. no Composition event). We created a relative timeline per available song, in order to provide an anchor point for automatically generated features. So far, this dump is linked to DBPedia (look at the event:place property on Performance events).

Another dump is the Jamendo one. It holds numerous links to external datasets, such as the Musicbrainz one, and the Geonames one. A more detailed presentation of how these links were created is available here.

Yet another dump it the John Peel Sessions one.

This material is Open Knowledge