W3C home > Mailing lists > Public > public-openannotation@w3.org > October 2014

Re: invalid anchors in OA specification

From: Stian Soiland-Reyes <soiland-reyes@cs.manchester.ac.uk>
Date: Fri, 31 Oct 2014 22:15:08 +0000
Message-ID: <CAPRnXtnsa2be8T23_gLyYZ=aqPSev-juVOu19yQU38S-uDznhQ@mail.gmail.com>
To: Robert Sanderson <azaroth42@gmail.com>
Cc: Nico Schlömer <nico.schloemer@gmail.com>, public-openannotation <public-openannotation@w3.org>, André Gaul <andre@gaul.io>
(Perhaps related)

The OA vocabulary HTML (as opposed to the spec/core that is the actual
spec) has been generated with LODE - you might see that

http://www.essepuntato.it/lode/http://www.w3.org/ns/oa

is very similar to http://www.w3.org/ns/oa.


We don't have much control of the outputs. In particular the anchor
links like #d4e119 is auto-generated every time and so for instance do
not work anymore at
http://www.essepuntato.it/lode/http://www.w3.org/ns/oa#d4e119

LODE does also have full-URI identifiers, like

  http://www.w3.org/ns/oa#http://www.w3.org/ns/oa#hasBody

(Why is it the full URI? Well, as far as LODE knows there could be
many hasBody from different namespaces within the same OWL).


.. but you just have to know about these - for some silly reason the
Table of Content is not linking to these and then not showing them to
a visitor.   (probably because it then can't distinguish between say
#hasBody the object property and #hasBody an annotation property that
is further down).



As a decision was made for oa.owl to use # instead of / - it is
impossible for the HTTP server to redirect to this "full" URI anchor.
For OA it should however be possible to do a quick search/replace in
the HTML so that the #http://www.w3.org/ns/oa#hasBody anchors instead
are just #hasBody.


I believe LODE is implemented as a mixture of Java and XSLT, and when
I tried to look at it, I struggled to find a good way to fix this
thing with the anchor links, as it just happens while iterating over
the XML elements. As a minimum I would want to get rid of the #a23948
thing as it meant non-stable (and non-readable) anchor links. The full
URI at anchor is a great ID -- although it is also technically invalid
URI to have an anchor that itself contains the # character..  (any
suggestion for an alternative character?)




In my never-present spare time, I'm working on an alternative, but
still similar way to do this kind of "spec style" rendering.

http://owl.s11.no/view/ont/http://www.w3.org/ns/oa.rdf

with a bit more control of the rendering and hyperlinks.

For instance I am planning a /respec/ style that uses respec.js for a
"W3C"-like rendering, e.g. MUST and SHOULD appear in colours. I am
also thinking to do add support for the prov:component annotations as
used in http://w3.org/ns/prov-o - for grouping related
properties/classes in sections.  (
http://owl.s11.no/view/ont/http://w3.org/ns/prov-o )


I have also added Markdown support, which means you get this nice mix
of being able to have HTML and/or Markdown in your rdfs:comment and
get it rendered somewhat sensible. Except for the odd hyperlink, I did
not find anything Markdown-like in the OA OWL now (like `code`,
*emphasis* or > quote).

Here is one example where it IS rendered as Markdown:

http://owl.s11.no/view/ont/http://purl.org/pav/#http://purl.org/pav/hasVersion


My OA approach is that I want to combine this with an annotation
server so that anyone can annotate any ontology, add examples,
questions, etc. These would most likely be relevant to any use of the
ontology, not just in the OWL Viewer. For instance they could also be
shown from within desktop tools like Protege after importing an
ontology.


I am however in my early days - as you might see I have not got as far
as getting subclasses or domain/ranges shown (as I was fighting myself
on how to find out if the link should be internal or external..).
"Pretty labels" for properties (e.g. "Comment" instead of
rdfs:comment) are not yet there. Also the 'main' properties should be
listed before arbitrary other annotations.




On 27 October 2014 16:01, Robert Sanderson <azaroth42@gmail.com> wrote:
>
> Hi Nico,
>
> The ontology terms probably should resolve to the right place in the
> representation, thanks for catching that.
>
> The specification itself, though, is at:
> http://www.openannotation.org/spec/core/
> in case you found the ontology document some other way.
>
> Rob
>
> On Tue, Oct 21, 2014 at 2:55 AM, Nico Schlömer <nico.schloemer@gmail.com>
> wrote:
>>
>> Hi,
>>
>> I'm looking at the Open Annotation Specification [1] and find that
>> many of the specified IRIs, e.g., [2], don't lead the to the
>> appropriate resource. The anchor corresponding to [2] is [3]. Is this
>> something that needs to be fixed?
>>
>> Cheers,
>> Nico
>>
>>
>> [1] http://www.w3.org/ns/oa
>> [2] http://www.w3.org/ns/oa#hasBody
>> [3] http://www.w3.org/ns/oa#d4e119
>>
>>
>
>
>
> --
> Rob Sanderson
> Technology Collaboration Facilitator
> Digital Library Systems and Services
> Stanford, CA 94305



-- 
Stian Soiland-Reyes, myGrid team
School of Computer Science
The University of Manchester
http://soiland-reyes.com/stian/work/ http://orcid.org/0000-0001-9842-9718
Received on Friday, 31 October 2014 22:15:57 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:22:06 UTC