I finally succeeded to find a server on which to publish the magnatune
dataset, available as linked data and through a SPARQL end-point.
I used D2R to publish it, and I built the mapping according to the
recent developments of the Music Ontology I will also add more links
to dbpedia and mbz relevant resources. [1] (however, it does not use
all the features of MO, as there are not a lot of things in magnatune
sql dump:-) ).
It is available there:
and the mapping is available at:
http://moustaki.org/magnatune/magnatune_mapping.n3
(Sorry for the name "dbtune", I did not have any ideas, so I just
stole the db* pattern of dbpedia:-) ).
However, I have run across a problem that some of you may have come
across as well. In the SQL dump of magnatune, there are information
about the country from which a track comes from, in the form of
literals such as "Ireland", "USA", "France", ... and I would like to
map these to the corresponding geonames features. But in order to do
so, I would need to plug some things into D2R (such as a map linking
the name of a country to its country code) - is there a not-too-hacky
way to deal with this sort of stuff?
I will try to add other creative commons licensed music repositories
online soon, at the same place.
Best regards,
Yves
[1] http://pingthesemanticweb.com/ontology/mo/
--
PhD student
Centre for Digital Music
Queen Mary, University of London
We have already computed some audio descriptors for all the Magnatune
collection (e.g tempo -beats per minute-, Key and mode, intensity,
danceability, etc.).
It's already in RDF (based on a simpler music ontology[1], the one
that foafing the music uses):
{rdf:Description rdf:about="http://he3.magnatune.com/all/20-Pimps-The
%20Strap%20Ons.mp3"}
{rdf:type rdf:resource="&music;Track"/}
{dc:title}Pimps{/dc:title}
...
{music:playedBy rdf:resource="http://magnatune.com/artists/
strapons" /}
{music:duration}T00H1M54S{/music:duration}
{music:key}B{/music:key}
{music:keyMode}minor{/music:keyMode}
{music:tonalness}0.68{/music:tonalness}
{music:tempo}96{/music:tempo}
{music:intensity}Energetic{/music:intensity}
...
{/rdf:Description}
Anyway, you can use these audio descriptors if you want to...
Regarding Artists and geolocation, I can provide you the lat and long
for most of the artists, too.
E.g.
{geo:Point geo:lat="39.766180" geo:long="-86.441050" /}
If you're interested we can discuss this off-line in more detail.
Cheers, Oscar.
[1] http://foafing-the-music.iua.upf.edu/ISWC2006/index.html
>
> Do you have a page somewhere describing the dataset and project?
>
I just put something online, it is available there:
http://purl.org/dbtune/doc/
I don't know if it is really useful so far though (I will explicit the
mapping and the structure of the magnatune SQL dump in details a bit
later).
Cheers!
Yves
First of all, great job Yves :)
MBZ will follow soon.
This would be quite interesting Oscar, please talk about it on the ML
so everybody will know what is going on with that, and/or you can
always reach us on #musicontology on Freenode.
Take care,
Fred