Re: shapes-ISSUE-91 (hsolbrig): Default Cardinality [SHACL Spec]

On 9/25/15 4:17 PM, Irene Polikoff wrote:
> Karen, I think you meant open shapes and closed shapes, not open and
> closed graphs which is not something the group has defined.
>
> Even without cardinality, there is typically some other constraint -
> that is a constraint on the value.

I don't know where you get "typically." I have a defined data schema 
with more than 500 fields and 1400 separate data elements, and they are 
either mandatory or optional. Since all but a few take uncontrolled 
text, there's not much to constrain.

I think we should quit talking about "typically." The world of metadata 
is very big and very diverse.

Further, one could specify only
> minCardinality or only maxCardinality, leaving the other one to
> default. With this, not specifying constraints for both min and max
> cardinality ( or even for neither) would not be meaningless from the
> data validation perspective - there are still things to check with
> open shapes.

My example was specifically with the defaults for cardinality within the 
open shape (which I think is a graph, but I don't care what we call it) 
method. That's what Harold's issue is about. Cardinality defaults in SHACL.

>
> I think there are two criteria to consider regarding default:
>
> 1. Intuitiveness
>
> To me, if there is no cardinality constraint stated, the intuitive
> interpretation is that there is no constraint.
>
> 2. Verbosity
>
> This, of course, will very much depend on the model, but we can look
> the commonly used vocabularies such as SKOS. Expressing SKOS in SHACL
> for constraint checking would be much more verbose if the default min
> 1, max 1. Majority of SKOS properties are min 0, max unlimited.

Can you cite your source for this? It actually doesn't make sense to me.

kc

>
> We could look it other models as well.  I find that proportion of
> properties that are {min 1, max 1} is much smaller than properties
> that are either {min 0, max unlimited} or {min 1, max unlimited} or
> {min 0, max 1 (or some other number)}. All of these would require
> more verbosity if the default was min 1, max 1.
>
> Sent from my iPhone
>
>> On Sep 25, 2015, at 8:15 AM, Karen Coyle <kcoyle@kcoyle.net>
>> wrote:
>>
>> I think that the cardinality defaults interact with the closed/open
>> graph definition. If the graph is open, then a default of
>> "minCardinality = 0, maxCardinality = *" is pretty close to
>> meaningless. In an open graph, all potential predicates are
>> "optional" unless defined otherwise, and specifying optional
>> predicates does not invoke any useful behavior. In the case of an
>> closed graph, "minCardinality = 0" describes a specific optional
>> predicate.
>>
>> SHACL, if I understand it correctly, describes an open graph by
>> default. This means that only ""minCardinality > 0" can be
>> validated.
>>
>> Although the statement by Holger that "if something is left
>> unspecified then it should count as unconstrained" resonates, I
>> would consider the inclusion of a optional property to be
>> specified, not unspecified.
>>
>> kc
>>
>>> On 9/25/15 1:02 AM, Holger Knublauch wrote: I believe if
>>> something is left unspecified then it should count as
>>> unconstrained. So if no sh:minCount or sh:maxCount is given then
>>> it should count as 0..* by default.
>>>
>>> PROPOSAL: Close ISSUE-91 stating that the default interpretations
>>> of sh:minCount and sh:maxCount (and their qualified counterparts)
>>> should remain as currently specified.
>>>
>>> Holger
>>>
>>> PS: A compact syntax may of course use different conventions and
>>> automatically generate the corresponding min/max triples.
>>>
>>>
>>>> On 9/25/2015 0:46, RDF Data Shapes Working Group Issue Tracker
>>>> wrote: shapes-ISSUE-91 (hsolbrig): Default Cardinality [SHACL
>>>> Spec]
>>>>
>>>> http://www.w3.org/2014/data-shapes/track/issues/91
>>>>
>>>> Raised by: Harold Solbrig On product: SHACL Spec
>>>>
>>>> The defaults for cardinality in UML are [1..1]  (see:
>>>> MultiplicityElement.lowerBound() and
>>>> MultiplicityElement.upperBound() on page 41 of OMG
>>>> specification ptc/2013-09-05).  Should these be the defaults
>>>> for mincount and maxcount in Section 3.1.5 of the SHACL
>>>> specification as well?  Currently they are [0..*].
>>
>> -- Karen Coyle kcoyle@kcoyle.net http://kcoyle.net m:
>> 1-510-435-8234 skype: kcoylenet/+1-510-984-3600
>>
>

-- 
Karen Coyle
kcoyle@kcoyle.net http://kcoyle.net
m: 1-510-435-8234
skype: kcoylenet/+1-510-984-3600

Received on Friday, 25 September 2015 15:13:57 UTC