Re: Proposal for ontology and api structure

Dear Felix, Silvia, all,

Nice progress. Regarding the API example, I agree with Silvia in having an 
example decoupled from HTML5. Maybe it is a naive idea, but why not making 
an API independent from the metadata embodiment (language and binding) and 
using just an URI to refer to the media object or fragment?

By the way, I've uploaded a preliminary version of the Digital Imaging 
Lifecycle UC to the wiki. I'm still working with Jaime and Victor on it, but 
this way you can check its state and give us some feedback.

Best regards,

Ruben


----- Original Message ----- 
From: "Silvia Pfeiffer" <silviapfeiffer1@gmail.com>
To: "Felix Sasaki" <fsasaki@w3.org>
Cc: <public-media-annotation@w3.org>
Sent: Monday, November 10, 2008 8:47 AM
Subject: Re: Proposal for ontology and api structure


>
> On Mon, Nov 10, 2008 at 5:30 PM, Felix Sasaki <fsasaki@w3.org> wrote:
>>
>> Hi Silvia, all,
>>
>> Silvia Pfeiffer さんは書きました:
>>>
>>> Hi Felix, all,
>>>
>>> Nice progress!
>>>
>>> I only have a brief comment on the API example.
>>>
>>> I would not attach the API example to the HTML5 video tag. That
>>> assumes that the annotation is associated to a video file or at least
>>> to the video tag in some way. I don't think you can assume that from
>>> the HTML5 standard or from a video file.
>>>
>>> Instead, I would define the API based on having a stand-alone
>>> annotation file, maybe a RDF file or something.
>>> And then I would encourage media file formats to encapsulate these
>>> annotation fields directly into the header of the video files and
>>> expose these to the video tag in a standard way. This standard way
>>> could be a javascript API - or maybe preferrably a DOM of its own.
>>>
>>> Just my thoughts on this. It is a difficult issue.
>>>
>>
>> Yes, it is. My impression currently is that we have very different 
>> opinions
>> on this topic. From the browser point of view, some people might even 
>> want
>> something like
>> Element vid = doc.getElementById("MyVid");
>> vid.getCreateDate();
>> that is, even closer alignment with the video tag. I'm not sure yet what 
>> the
>> way out is here.
>
> That would require the video tag to expose a javascript API with that
> functionality. But where does the video tag get that information from?
> It can either come from within the video file (where in turn you
> require an API towards the video) or it comes from an external file
> that is related to the video (e.g. a RDF file) through another
> attribute.
>
> I agree that such a functionality would be nice to have. But it's up
> to the HTML5 WG to define this API, really. All we can define here is
> the API to the annotation file or propose an API into the video
> formats that include the annotation tags inside themselves.
>
> That's all I was thinking.
>
> Cheers,
> Silvia.
>
>

Received on Monday, 10 November 2008 11:02:47 UTC