AW: top tag descriptions, clarifications

Dear Felix,

I fully agree that we should collect this information in the next round of updates of the mapping table. However, as David pointed out, this wil require to define it for our set of properties.

Best regards,
Werner

Von: public-media-annotation-request@w3.org [mailto:public-media-annotation-request@w3.org] Im Auftrag von Felix Sasaki
Gesendet: Donnerstag, 21. Mai 2009 19:07
An: Pierre-Antoine Champin
Cc: David Singer; public-media-annotation@w3.org
Betreff: Re: top tag descriptions, clarifications

Hi Pierre-Antoine,

no need to repeat previous discussions about return values ... my concern was not about the API, but about information in the mapping table, which is not available yet.

Felix
2009/5/21 Pierre-Antoine Champin <pierre-antoine.champin@liris.cnrs.fr<mailto:pierre-antoine.champin@liris.cnrs.fr>>
A few comments:

- I think we should not distinguish URN from dereferenceable URIs; this
debate is out of scope for us. Any URI should be considered as an
identifier, possibly dereferenceable.

- I think we should allow textual descriptions or URIs, but also be able
to disinguish between the two

- I think we should allow explicit multiple values

The last two points have been my concerns when developing the toy
application at [1] (which should be updated now with the top tags).

 pa

[1] http://liris.cnrs.fr/~pchampin/wsgi/mawg/<http://liris.cnrs.fr/%7Epchampin/wsgi/mawg/>

David Singer a écrit :
> I have an action to improve the descriptions
> on<http://www.w3.org/2008/WebVideo/Annotations/wiki/Top_Supported_Tags>
>
> I'm not quite sure what sort of improvement I should be allowed to do,
> however.
>
> Let's see where I think an improvement might be indicated.
>
> For many tags:
>   -- in what language is this expressed?  Are multiple e.g. titles in
> different languages allowed?
>
> Enumerations:
>   -- for tags like contributor (contribution role), genre etc.: are we
> planning to have a defined set of names, some extension capability, or a
> free-for-all?
>
> For names
>   -- do we allow a de-referencable pointer as well as a textual name?
> E.g. are <http://www.sony.com> and "Sony Corporation" permitted?  How do
> we distinguish the two, if so?
>
> For couplets (contributor, relation):
>   -- we need to say how the pairing is done.  Is this textual, e.g. {a, b}?
>
>
>
> ma:creator -- is this the 'human names' of the creator(s), or their
> unique IDs (e.g. URN) or  de-referencable pointer (e.g. URI to their
> FaceBook page :-))?
>
> ma:keyword -- if there is more than one word, is it a phrase, or are
> they separate keywords?  Do we need to say phrases are in quotes? Are
> keywords an 'enumerated space' (as above)?
>
> ma:duration -- someone at the meeting was arguing that this might be
> different from the actual media duration, but I can't remember why. If
> it's the actual media duration, it's a property rather than annotation.
>
> ma:spatialDimension -- we need units, and we need to know if they are
> uniform.  I would say that the dimensions must be uniform (square grid)
> and that some unit must be specified (cm, inch, square-pixel, point, ...)
>
> ma:genre -- enumeration?  (note typo in the 4th column, 'resrouce')
>
> ma:createDate -- can we say that this is a date as defined in ISO 8601
> string format?  That allows for presentation, comparison, conversion to
> other date systems etc.  We probably don't need to care about the b.c.e
> issue.
>
> ma:compression -- what namespace is used to name compressors?  Or is
> this a URI to the spec.?
>
> ma:location -- probably an ISO 6709 (Annex H) string, right?  Do we want
> other possible additional values here (e.g. a human name, "Barcelona")
> etc.?
>
> ma:collection -- rather than a name, wouldn't an identifier (URN or URI)
> be better?
>
> ma:rating -- how is the person or organization identified?
>
> ma:targetAudience -- isn't this a triplet?  e.g. "Venezuelan Board of TV
> Censors", "Permissible broadcast hours", "03h - 04h"?  Why is rating not
> one of these triples?
>
>