Jeremy Carroll wrote:
Cc-ing ivan and bijan since they seem interested.
I am going to start a variety of threads on www-archive, with the intent
that we pick the threads up on the tuesday meeting.
I take it that the theme of the discussion is trying to evaluate powder
design options to get appropriate balance between 'correctness' with
respect to the more mathematical bits of the SW Recs (OWL and RDF), and
usability by the intended user base.
That is the essential tension we are trying to resolve, yes.
The ideal outcome would be one where small amounts of attention to
detail, make little to no impact on usability, but enhance 'correctness'
and hence interoperability with off-the-shelf SW tools.
Indeed. As you have discerned, Jeremy, usability is a high priority
since we hope that POWDER will be of use well beyond the traditional
Semantic Web Community and therefore by some people for whom the
mathematical exactness of the RDF/OWL model will of little immediate
concern.
Poor outcomes would be ones in which tension between the two objectives
(correctness and usability) become conflict; and either POWDER users
feel that usability is sacrificed for ill-defined and elitist logical
goals, or the SW specs, at least their logical parts, are simply ignored
because they are unhelpful.
Yep.
I'll try and start these threads:
- reification, or what?
I dislike reification, and will argue against it, and in favour of
the design in the powder-dr WD, with minor mods.
I find this both pleasing and surprising. The POWDER WG has been working
with the structure in the currently published WD [1] for a good length
of time and we're comfortable with it as an image in our heads -
_however_ we do not, of course, want to promote something that is
clearly 'wrong' in an SW sense. So if we could fix the model with minor
modifications to [1] - I think we'd be smiling all the way.
- resource descriptions, as a semantic extension??
RDF semantics makes no provision for examining the IRIs used for
resources. This is not addressed by OWL 1.0 or OWL 1.1. This is
fundamental to POWDER; and should, in my view, be addressed by formally
creating a suitable semantic extension.
Sounds good to me.
- resource descriptions and monotonicity
I got a bad non-monotonic feeling while reading the powder-grouping
WD; interestingly it was while reading bits that had clearly been
written with the issue in mind :(
From my perspective it seems that the problem that jumps out at SW folk
in the grouping of resources doc [2] makes them jump to the
OWL-based/sub class of approach as exemplified in [3] and graphed at [4]
. But this then introduces reification which makes people run back the
other way!
Actually, I played with the reification issue a little today - and
_didn't_ put an ID on the subClassOf triple, i.e. the assertion that A
is a subclass if B is only in the reification triple - which gives the
RDF at [5] and the graph at [6] - which I have to say doesn't look too
bad to my inexpert eye - but I'm not an inference engine.
- the subclass relationship
I take the reified rdfs:subClassOf triple in the example that has
circulated in earlier e-mail to be a response to a request to be more
formal, or more OWL like, in response to the powder-dr WD. However, the
design in the WD looks more user-friendly to me, if it could be made to
work ...
- tools and motivations
Why o why are we bothered?
Aye, there's the rub Mr. C.
Phil.
[1] http://www.w3.org/TR/2007/WD-powder-dr-20070925/#structure
[2] http://www.w3.org/TR/2007/WD-powder-grouping-20071031/
[3] http://www.fosi.org/projects/powder/example4.owl
[4] http://www.w3.org/RDF/Validator/ARPServlet.tmp/servlet_14339.png
[5] http://www.fosi.org/projects/powder/example5.owl
[6] http://www.fosi.org/projects/powder/example5.png
--
Phil Archer
Family Online Safety Institute
w. http://www.fosi.org/people/philarcher/
