Re: using classes to control constraints

On Feb 7, 2015 5:37 PM, "Peter F. Patel-Schneider" <pfpschneider@gmail.com> 
wrote:
>
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
>
>
> On 02/07/2015 07:05 AM, Eric Prud'hommeaux wrote:
> > * Peter F. Patel-Schneider <pfpschneider@gmail.com> [2015-02-07
> > 06:44-0800]
> >> In a discussion about the LDOM Primer on 02/07/2015 01:58 AM, Eric
> >> Prud'hommeaux wrote: [...]
> >>> "This instance passes/fails this shape" is quite clear. Adding a type
> >>> arc is effectively a non-starter for this group; there are too many
> >>> people who see that is hampering re-usability of the data.
> >>
> >> Do you mean to say that there is no chance that any prominent example
> >> of constraints working off types will pass muster?
> >>
> >>
> >> I am very strongly in favour of having shapes be different from RDFS
> >> classes but I also very strongly believe that a common situation is
> >> that constraints are triggered from class membership.  This common
> >> situation should be prominent in the working group's documents.
> >
> > I'm skeptical that it's a common occurance in sensible modeling, but I'm
> > certainly happy to be shown otherwise.
>
> I expect that class attachment is the most common trigger for constraints 
in
> SPIN, but maybe the people from TopQuadrant have some usage figures that
> would verify this expectation.
>
> > Its possible that our disagreement stems from different starting
> > conditions. Here are mine:
> >
> > Much of the value of RDF stems from "serendipitous reuse".
>
> Agreed.
>
> > The prominent examples should use the core shapes language.
>
> I agree that the prominent examples should use the core portion whatever 
is
> produced from the WG.
>
> > Physical laws like area aren't typical of business logic.
>
> Agreed, but this doesn't seem to be relevant to the issue at hand.
>
>
>
> Here is what I would expect from the first example in a primer.
>
> 1/ There is an ontology for something.  (Let's use bugs.)   The ontology
> describes several classes, such as bugs and people, and several 
properties,
> such as state and reporter, all in an open-world setting.
> 
> 2/ There is some RDF data that uses this ontology, describing one or more
> bugs and people.
>
> 3/ There is some wording that introduces the notion of verifying that
> sufficient information is present so that useful things can be done with 
the
> RDF data.

I think I can address this with a "record class" as described in
<http://www.w3.org/2015/02/shapes-article/> (many thanks for your
feedback on that document).


> 4/ A set of constraints/shapes are given whose effect is that if the data
> correctly validates the bug instances do indeed have sufficient 
information.
>  These constraints/shapes are triggered off the bug class.

http://w3c.github.io/data-shapes/data-shapes-primer/no-class-templates.html#associations
now includes a whole slew of associations with shapes, the first of
which is ldom:instanceShape, second is ldom:classShape (editorially
made sense in that order).

[[
clinic:PatientRecord a owl:Class ;
  ldom:classShape [
    ldom:property [
        ldom:predicate clinic:phone ;
        ldom:valueType xsd:string ;
        ldom:minCount 1 ; ldom:maxCount 1
    ]
  ] .
]]

Is that good enough for an FPWD to tell the world what we're up to?


> I believe that this example satisfies all your desiderata above.
>
> peter
>
>
>
>
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1
>
> iQEcBAEBAgAGBQJU1j8+AAoJECjN6+QThfjzlzcH/0BT7Ld8qtf/KxD9cbnghNrX
> cyJqJwp+oWPyKWH60xABz2rmecY4e027EKOnkZutE05z1rzDcaFudWOpnRB6O5vD
> tgp2wUaDi63+4wJx5nnvqZMoWxPhfssy/zS8WSAJYveE9AQujQNdrusTgV5mOEh5
> 7xSsaovdFB/o+/jFNzv0KXRtfCG2JCR9TUZtg0KKibHCR1mGXmvaCcspYcrEGC9t
> jLUnokS513KsCmtF/IccuoJYX6+S7xUd43MqYXpqYgLrKs7aa6XqYzIPnlSGcZL6
> tyVAoMR3F/jaMDZjMsW65vyKqIvMlQfipsAWMPpmb8OFooBsLcyjkSehLzJoFCU=
> =khFQ
> -----END PGP SIGNATURE-----