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

Re: Selection Filtering

From: James Smith <jgsmith@gmail.com>
Date: Fri, 10 Aug 2012 08:21:52 -0400
Cc: Robert Sanderson <azaroth42@gmail.com>, Paolo Ciccarese <paolo.ciccarese@gmail.com>, public-openannotation <public-openannotation@w3.org>
Message-Id: <5251E758-3855-4DD2-945E-3B8A2F831420@gmail.com>
To: Stian Soiland-Reyes <soiland-reyes@cs.manchester.ac.uk>
There are two different cases here: (1) the target is embedded within a resource and is not directly addressable (but the target can be specified exactly by current OA and OAX properties), (2) the target has its own existence, but the annotation only applies when that target is in the context of another target (something that can't be specified with current OA or OAX properties).

The second case is what oax:hasContext aims to provide.

For example, if I want to annotate a particular paragraph in a page, that paragraph is not addressable itself, but must be selected out of the page. I'd use an XPath or similar to select the paragraph and have the page be the oa:hasSource for the oa:SpecificTarget. There's no need for oax:hasContext.

If I want to annotate the MITH logo on http://mith.umd.edu/, but not on http://mith.umd.edu/about/ (or anywhere else the logo appears), then I'd construct the annotation as OA would have us already do, targeting the logo, but add a oax:hasContext pointing to the http://mith.umd.edu/ indicating that the annotation only applies when the logo is in the http://mith.umd.edu/ context, whatever that might be (in this case, an HTML page referencing it through an <img/> element). I'm not commenting on the <img/> element, but on the resource it points to, and the annotation only makes sense when that resource is considered in the context of that particular page.

Another way to think about it is how to select annotations of objects referenced in a web page, and only selecting those annotations that were made about that object when it was presented as part of that web page, but also being able to do analytics about all annotations about that object regardless of where the object was seen when the annotations were made.

-- Jim


On Aug 10, 2012, at 4:57 AM, Stian Soiland-Reyes <soiland-reyes@cs.manchester.ac.uk> wrote:

> On Thu, Aug 9, 2012 at 11:13 PM, Robert Sanderson <azaroth42@gmail.com> wrote:
>> Does the hasContext approach (eg climbing up rather than drilling
>> down) deal with situations when there's really embedded content, or
>> only resources that are referenced but have their own unique URIs?
>> 
>> For example, the use case I have in mind is an ePub document
>> (basically a zip file that contains HTML and related content) that has
>> a URI, but the chapter xhtml files within it do not.  And similarly
>> the images referenced from those chapters don't have their own URIs,
>> they're just named bitstreams within the zip.
>> 
>> I don't immediately see it, if it does.  And if not, how would we go
>> about providing a solution for the use case?
> 
> They *could* have a URI constructed for them, but it would be
> something quite specific, similar to Java's jar: protocol (which
> actually would work with ePubs, as they are ZIPs):
> 
> jar:http://example.com/ebook.epub!/path/within/archive/page.html
> jar:http://example.com/ebook.epub!/path/within/archive/picture.jpeg
> 
> 
> But I take your argument; there are many compound resources where we
> want to refer to things inside it, and then things inside that inner
> compound resource again.
> 
> 
> So why not just a chain of specific resources?
> 
> :aboutPictureOnPage a oa:Annotation ;
>  oa:hasTarget :pictureOnPage .
> 
> :pictureOnPage a oa:SpecificResource ;
>  oa:hasSelector :somePictureSelector ;
>  oa:hasSource :page .
> 
> :page a oa:SpecificResource ;
>  oa:hasSelector :somePageSelector ;
>  oa:hasSource <http://example.com/ebook.epub> .
> 
> I am not too keen on this, because here oa:hasSource suddenly goes to
> an 'abstract' resource, and given <http://example.com/ebook.epub> it
> would be much harder to find annotations on anything inside it.  This
> could however be sane when a specific resource also has a URL (which
> now that we moved hasStyle is kind of OK), and you can't decide which
> is the top level:
> 
> 
> :aboutPictureOnPage a oa:Annotation ;
>  oa:hasTarget <http://example.com/ebook/page25/image.jpeg> .
> 
> <http://example.com/ebook/page25/image.jpeg> a oa:SpecificResource ;
>  oa:hasSelector :somePictureSelector ;
>  oa:hasSource <http://example.com/ebook/page25/> .
> 
> <http://example.com/ebook/page25/> a oa:SpecificResource ;
>  oa:hasSelector :somePageSelector ;
>  oa:hasSource <http://example.com/ebook/> .
> 
> 
> 
> 
> 
> I thought I would prefer the oax:hasContext (we can make it allowed to
> be chained to other oa:SpecificResource's) - here I try to make an
> example of multiple layers:
> 
> Multi-layer contexts:
> 
> :aboutAreaOfPictureOnPage a oa:Annotation ;
>  oa:hasTarget :areaOfPictureOnPage .
> 
> :areaOfPictureOnPage a oa:SpecificResource ;
>  oa:hasSelector :someSvgSelector ;
>  oax:hasContext :pictureOnPage ;
>  oa:hasSource <http://example.com/ebook.epub> .
> 
> 
> :pictureOnPage a oa:SpecificResource ;
>  oa:hasSelector :somePictureSelector ;
>  oax:hasContext :page ;
>  oa:hasSource <http://example.com/ebook.epub> .
> 
> :page a oa:SpecificResource ;
>  oa:hasSelector :somePageSelector ;
>  oa:hasSource <http://example.com/ebook.epub> .
> 
> 
> This has some problems. For instance :areaOfPictureOnPage might look
> like just running an SVG Selector on the epub archive. This would not
> work - but for some combinations of selectors and source it could look
> like it works, like an xpath selector on the XHTML embedded as text
> within an XML which is also selected on with xpath.
> 
> So for the hasContext approach, we would also need to add something
> like a oa:hasState [ a oax:ContextState ]  to prevent that
> interpretation.  Then the oax:hasContext could be stated in there
> instead of in the oa:SpecificResource.
> 
> 
> -- 
> Stian Soiland-Reyes, myGrid team
> School of Computer Science
> The University of Manchester
Received on Friday, 10 August 2012 12:22:23 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 10 August 2012 12:22:23 GMT