W3C home > Mailing lists > Public > public-openannotation@w3.org > July 2012

Re: [Open Annotation] some questions

From: Robert Sanderson <azaroth42@gmail.com>
Date: Fri, 20 Jul 2012 11:35:02 -0400
Message-ID: <CABevsUG+rtqruRy=qO3aM_5=rOstNq4xdMwjDyE3=s_qnFeAxA@mail.gmail.com>
To: Christian Morbidoni <christian.morbidoni@gmail.com>
Cc: Paolo Ciccarese <paolo.ciccarese@gmail.com>, public-openannotation@w3.org
On Thu, Jul 19, 2012 at 9:33 AM, Christian Morbidoni
<christian.morbidoni@gmail.com> wrote:
> 1) When I use pundit:hasPageContext, I think about a property attached to
> the annotation and not to one of the specific targets.
> In Pundit we are supporting cross-page annotations, that is an annotation
> that has two targets, e.g. two named content from two different pages.

> Then...again: how do I know what page the user was looking at when he
> created the annotation? was ithttp://example.org/page/2.html or
> http://example.org/page/1.html

I'm not sure I see the requirement.  Can you explain how a consuming
application would change its behaviour based on the information as to
which page the user was looking at when she created (submitted?
started?) the annotation with multiple targets?

> That is why I'm currently doing:
> <_:Anno1> a oa:Annotation ;
>   oa:hasSource <_:Body1> ;
>   oa:hasTarget <_:SpecificResource1> ;
>   oa:hasTarget <_:SpecificResource2> ;
>   pundit:hasPageContext <http://example.org/page/1.html> .

To be honest, I think that's fine! It seems like a particular piece of
information that your client can create and then use, without
affecting anything else.  Other systems will simply ignore the
hasPageContext relationship unless they also know what it means.

> 2) You are right when you say that there is probably no need for the class
> oax:AboutAttributeSelector
> Also I think it would be better to use the URL of the named content as value
> of the hasSource property (e.g.http://example.org/section/123) as it is a
> web resource and should have a HTML representation itself (e.g. a HTML page
> including only one paragraph).

And thus Paolo's interpretation of X in the context of Y.  I blame
jet-lag for my misunderstanding, my apologies for being dim! :)

We discussed this early on and filtering of annotation targets and
bodies seems to open quite a large can of worms.  For example,
annotating resource X as it appears in resource Y is just a specific
case of all instances of X, with some added restriction.   We've tried
to limit the scope of annotations to exactly one "thing" per target (a
single span of text, a single area in an image, a single timeslice in
a video etc) rather than a named set of, essentially, search results.

Personally, I'm nervous of the effect of allowing sets of things like
this.  One could simply annotate /html/body without specifying a
target HTML page and clients would naively render it *everywhere*. Or
occurrences of "the" in wikipedia. Or ...

I realize that's a far cry from what is being discussed, but I think
that any solution for the current issue would need to make it
impossible for this sort of thing to occur.

Does that make sense?

Received on Friday, 20 July 2012 15:35:35 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:38:10 UTC