What about music-related URIs???

[Kurt J]

Hello All,

It seems there are (too) many options for creating music-related URIs.
They are mostly all based on Musicbrainz.org IDs (which is quite
appropriate), but it can be difficult to know which to choose. I can
identify at least four different sources of artist URIs based on
musicbrainz ids:

1. dbtune.org/musicbrainz
pros: there are uris for records, tracks, and artists (and
more), they dereference nicely; a nice comprehensive mapping
cons: it's updated only periodically from the musicbrainz dumps

2. bbc.co.uk/music
pros: well maintained by hard-working professionals; deref
nicely w/ fancy content neg
cons: afaik, no URIs for releases, tracks - just artists; it is
arguably biased by bbc interests (not saying i believe this myself)

3. zitgist.com
pros: samesAs links from DBpedia; includes URIs for records /
tracks / etc
cons: not sure how often they are updated from musicbrainz;
not sure why, but seem to have fallen out of fashion somewhat (any
reasons?)

4. musicbrainz.org
pros: this is 'the source' of all these ids and suc
cons: afaik, these do not actually deref to any RDF;

also note, URIs like this

http://musicbrainz.org/music/artist/20072862-5132-478e-b9e2-8a3aa308cd89

appear in http://lod.openlinksw.com/

i think this might be a mistake in xslt or something - not only do
these not actually deref to rdf, they return 404 from Musicbrainz -
perhaps there is some thing i am not understandin tho. i believe the
'/music/' should be removed - then musibrainz.org handles the URI and
returns an html page.

As I mentioned before, we are giving a tutorial on music information
retrieval and the semantic web, see:

http://ismir2009.dbtune.org/

Put simply, what URIs do I tell people to use?

Thanks.

-Kurt J

[Nicholas J Humfrey]

Hello,

It is a dilemma indeed. I think we need to make a effort to get RDF/
RDFa on to musicbrainz.org and then we can use the 'proper' URIs. I
have been holding back from doing this because the codebase has been
changing so much - but there is a big new release coming soon....

nick.

[Kurt J]

Hi Nick,

> It is a dilemma indeed. I think we need to make a effort to get RDF/
> RDFa on to musicbrainz.org and then we can use the 'proper' URIs. I
> have been holding back from doing this because the codebase has been
> changing so much - but there is a big new release coming soon....

i like using the actual musicbrainz URIs as well. to be clear, the
artist URI would be of the form

http://musicbrainz.org/artist/<mbid number>

similarly the track URI would be

http://musicbrainz.org/track/<mbid number>

and the release (record) URI would be

http://musicbrainz.org/release/<mbid number>

i've noticed that currently the release-based URIs return 404 unless
you put '.html' on the end

http://musicbrainz.org/release/3de4b060-3930-4d6f-8c4f-7135b7d70fb6 -> 404

http://musicbrainz.org/release/3de4b060-3930-4d6f-8c4f-7135b7d70fb6.html
-> page about this release

track and artist URIs seem to redirect sensibly to a human-readable page.

So do you advocate using the Musicbrainz 'proper' URIs now? Any guess
on how soon they might include RDF(a) ?

-Kurt J

[Yves Raimond]

I guess it will be as soon as we actually send them a patch :-) I did
some work on that last year (just rdf/xml) but as nick said, the
codebase has been changing very fast since then... We should gather
efforts once the big scary release is out and make it happen!

Cheers,
Y

[Dan Brickley]

On Sun, Sep 6, 2009 at 6:01 PM, Yves Raimond<yves.r...@gmail.com> wrote:
>
> I guess it will be as soon as we actually send them a  patch :-) I did
> some work on that last year (just rdf/xml) but as nick said, the
> codebase has been changing very fast since then... We should gather
> efforts  once the big scary release is out and make it happen!

It would be great to have the core music-brainz derrived music URIs
hosted on a *.musicbrainz.org domain, even if initially it served as a
redirector / mirror / cache for materials elsewhere. MB deserves to be
the hub here I think, and changing all those URIs in a couple of years
will be quite tricky... maybe some DNS based delegation is the way to
go in the short term?

cheers,

Dan

ps. i am just running MusicBrainz Picard tagger over all my mp3s
(includes all the CDs I ever bought...). ... Is there a tool /
workflow people here can recommend to turn all that into RDF locally?
does picard keep a db somewhere, or all metadata is just within the
files?

[Yves Raimond]

>
> ps. i am just running MusicBrainz Picard tagger over all my mp3s
> (includes all the CDs I ever bought...). ... Is there a tool /
> workflow people here can recommend to turn all that into RDF locally?
> does picard keep a db somewhere, or all metadata is just within the
> files?

You can try GNAT, in the motools project on sourceforge. It will
generate RDF linking your music collection to Musicbrainz identifiers.

Cheers,
y

>
> >
>

[Kingsley Idehen]

All,

Using the native MBZ URIs within the MBZ domain is the way forward.
Naturally, the pages should also include RDFa and then they won't need
to redirect anything. Basically, stick with HTML+RDFa as the default
metadata representation served up from data space: musicbrainz.org . It
is going to be much easier for User Agents to become RDFa aware than it
would ever be for Web publishers to globally implement Content
Negotiation based Identity dissambiguation. Of course, this doesn't
invalidate the deep utility of Content Negotiation -- in general -- it
simply means that its relationship to Linked Data Web boostrap
(activation threshold reduction) is inverse in nature :-)

You must be able to publish rich metadata in text form without
possessing Web Server admin privileges, bottom line.

MBZ could be another powerful boostrap point for the Web of Linked Data.

--


Regards,

Kingsley Idehen Weblog: http://www.openlinksw.com/blog/~kidehen
President & CEO
OpenLink Software Web: http://www.openlinksw.com

[Kurt J]

so those who have responded seem to agree using musicbrainz.org URIs
is the best. so i purpose we encourage the use of these in all future
/ current projects. then after this next mbz release work to inject
RDFa asap.

i think one thing that would really facilitate this is including
sameAs in DBpedia to musicbrainz artists along side the zitgist links.
I would suspect DBpedia is many peoples first entry point to LOD. but
perhaps this is another thread for another list ;-)

[Kingsley Idehen]

Kurt J wrote:
> so those who have responded seem to agree using musicbrainz.org URIs
> is the best. so i purpose we encourage the use of these in all future
> / current projects. then after this next mbz release work to inject
> RDFa asap.
>
> i think one thing that would really facilitate this is including
> sameAs in DBpedia to musicbrainz artists along side the zitgist links.
>

Drop the Zitgist links. The BBC corpus should be more extensive, and If
not, I'll have another dump produced (we've produced the dumps behind
various data space URIs to date).

> I would suspect DBpedia is many peoples first entry point to LOD. but
> perhaps this is another thread for another list ;-)
>

DBpdia should have a more recent dump than the initial Zitgist dumps.

Kingsley

[Kingsley Idehen]

David Peterson wrote:

>
>
> On 7/09/2009 5:44 AM, Kingsley Idehen wrote:
>> Using the native MBZ URIs within the MBZ domain is the way forward.
>> Naturally, the pages should also include RDFa and then they won't
>> need to redirect anything. Basically, stick with HTML+RDFa as the
>> default metadata representation served up from data space:
>> musicbrainz.org . It is going to be much easier for User Agents to
>> become RDFa aware than it would ever be for Web publishers to
>> globally implement Content Negotiation based Identity
>> dissambiguation. Of course, this doesn't invalidate the deep utility
>> of Content Negotiation -- in general -- it simply means that its
>> relationship to Linked Data Web boostrap (activation threshold
>> reduction) is inverse in nature :-)
>>
>> You must be able to publish rich metadata in text form without
>> possessing Web Server admin privileges, bottom line.

> +1
>
> Now if the HTML 5 mob just saw it that way... ;) Now that really is
> the topic for another conversation!
>
> I am oh so happy to be hearing this more and more. Embedding RDFa into
> HTML is the way forward for the vast majority of websites. It removes
> the whole complexity of server configs, content negotiation, what is
> and isn't a thing, blah, blah blah.
Yep! Esp. the "what is and isn't a thing.." odd discourse. As I've
stated a zillion times, Identity models aren't new to computer science
or the computer industry at large, they predate all of us :-)
>
> I worked with the ABC (Australian Broadcasting Corporation) to launch
> a digital radio site built with Drupal. We were faced with what URI to
> use and this conversation will indeed help. I will update the code
> accordingly to musicbrainz.

Great!

Kingsley
>
>
> Cheers,
>
> David