W3C home > Mailing lists > Public > public-media-annotation@w3.org > June 2009

Re: [mawg] Re: Audio and music?

From: Pierre-Antoine Champin <pierre-antoine.champin@liris.cnrs.fr>
Date: Tue, 23 Jun 2009 13:37:05 +0100
Message-ID: <4A40CC71.6020404@liris.cnrs.fr>
To: Yves Raimond <yves.raimond@gmail.com>
CC: Felix Sasaki <felix.sasaki@fh-potsdam.de>, public-media-annotation@w3.org

I attempted to address this kind of problem in my (now outdated, I 
should use the ma: properties now) toy implementation.


The same kind of problem exists in ID3, where some fields apply to the 
encoding, others to the recording, others to the performance.

So my implementation implicitly defines other resources, producing 
something that can be expressed in N3 as follow:

     dc:coverage "2.483" ;
     dc:date "2009-02-09" ; # from TDEN
     dc:source <#mawg-id3:recording> .

     dc:date "2009-02-09" ; # from TDRC
     dc:source <#mawg-id3:performance> .

     dc:date "2009-02-10" ; # from TDRL
     dc:creator "Pierre-Antoine Champin", "John Doe" ; # from TPE1
     dc:contributor "P-A. Champin" . # from TCOM

Unless you check "strict", the implementation will try to find the 
"closest" (inherited) value for the required property, so the 
dc:contributor of <> will be "P-A. Champin".

If you check "multiple", all possible values (inherited or not) will be 
returned (unless again "strict" is checked).

If you check "structured", you will get more metadata about the values, 
including a flag indicating whether the value has been inherited or not.

This is just a personnal proposal, but I think it is a good general 
principle of dealing with this complexity while keeping simple usage 
scenarios simple for the API user.


In my example, checking "multiple" and "structured" allows you to see

Le 23/06/2009 10:08, Yves Raimond a écrit :
> Hello!
>>>>> I just came across the following table:
>>>>> http://www.w3.org/2008/WebVideo/Annotations/wiki/FeaturesTable
>>>>> which (I hope, otherwise just ignore this email) is maintained by this
>>>>> group.
>>>>> Does the scope of the ontology you aim at developing also encompass
>>>>> audio and music? I was wondering at that when I saw ID3 mentioned in
>>>>> the table.
>>>> The main focus of our group is video. We are also looking into other
>>>> multimedia vocabularies, including ID3. However, I cannot tell (yet) to
>>>> what
>>>> extend these will be taken into account.
>>> First of all, congratulations on the public draft!
>> Thank you for the flowers :)
>>> At this occasion, I just took a look at the latest mapping table, and
>>> a few audio or music-related format are taken into account - thanks :)
>>> When reading the table, I noticed that the mapping to your core
>>> properties from existing formats worked really well for flat schemas
>>> (e.g. DC or ID3), but the explanations of the mapping for other
>>> (non-flat) schemas were missing. For example, ma:keyword is mapped to
>>> adaptationOf in FRBR, and ma:title is mapped to realizationOf /
>>> embodymentOf / exemplarOf. It is quite hard to guess (even for
>>> FRBR-savvy people) what that could mean, and how we could proceed to
>>> write a mapper from a FRBR-based vocabulary (e.g. Music Ontology) to
>>> the Ontology for Media Resource.
>> I completely understand your problem. To be able to solve it, some test data
>> would be very helpful. Could you send us some prototypical test data you
>> would use for your mapper?
> Yes, of course! Here is the RDF snippet I used (stripping out the
> unneeded stuff wrt. ma:title)
> @prefix mo:<http://purl.org/ontology/mo/>  .
> @prefix owl:<http://www.w3.org/2002/07/owl#>  .
> @prefix dc:<http://purl.org/dc/elements/1.1/>  .
> @prefix :<#>  .
> :work
>      a mo:MusicalWork ;
>      dc:title "Franz Schubert's Trout Quintet" .
> :performance
>      a mo:Performance ;
>      dc:title "Trout Quintet, performed by the LSO" ;
>      mo:performance_of :work ;
>      mo:recorded_as [
>          a mo:Signal ;
>          mo:published_as :track1 ;
>          dc:title "Recording of the LSO performing the Trout Quintet" ;
>      ] .
> :track1
>      a mo:Track ;
>      mo:track_number 5 ;
>      owl:sameAs<http://dbtune.org/musicbrainz/resource/track/3208fbce-c20f-4362-a3d5-5405ac1904bd>
> ;
>      dc:title "Trout Quintet, performed by the LSO, on 'Favorite Classics'" .
> So as you can see, this RDF describes a particular track in my
> collection, and holds three dc:title. It looks like the mapping table,
> for FRBR (on which this snippet is based), suggests that I should use
> a specific level of abstraction (the work, the performance or the
> track) to generate ma:title, but the mapping just says "realizationOf,
> embodimentOf  and exemplarOf".
> I hope that helps!
> Cheers,
> y
Received on Tuesday, 23 June 2009 12:38:04 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 23 June 2009 12:38:05 GMT