Re: Shape Selectors

No, the shape selectors section is limited to the discussion on how to connect shapes to classes.

The LDOM primer talks about global constraints, etc. If you are not interested in class-specific constraints, you don't need to read the shape selectors section.

Irene

Sent from my iPhone

> On Feb 13, 2015, at 9:59 AM, Karen Coyle <kcoyle@kcoyle.net> wrote:
> 
> 
> 
>> On 2/12/15 4:49 PM, Holger Knublauch wrote:
>> Hi Karen,
>> 
>> Shape selectors are entirely about addressing how to connect a shape to
>> a class for the use cases where a connection is needed. Thus, reading
>> about them will not give you any information on how to use LDOM to
>> define constraints that are not class-specific.
> 
> So LDOM is limited to class-specific constraints? I thought we were discussing how to connect shapes to instance data, not how to implement class-based shapes. Perhaps that is my confusion. I thought that we had already agreed that there were "global" constraints. Those don't appear here. And I have given examples of data that uses no declared classes but still has constraints, as well as graphs that are multiple classes. Perhaps what we need are more examples of instance data.
> 
> We don't seem to get ahead here. As Peter said on the call, we are in the weeds, because there are still fundamental problems that we do not or cannot address.
> 
> kc
> 
> 
>> 
>> Constraints  and data validation is a complex technical topic. Similar
>> in complexity to SPARQL and OWL specifications. I have tried to be as
>> clear as possible. There is the LDOM model. There are examples.
>> 
>> It would help if you pointed out paragraphs and sections in SPARQL and
>> OWL documentation (since they are comparable in terms of the technical
>> complexity) that make their documentation more accessible to you than
>> the LDOM primer. This would give me some ideas on how to best improve
>> the document to address your needs.
>> 
>> Regards,
>> Holger
>> 
>> 
>>> On 2/13/2015 1:03, Karen Coyle wrote:
>>> I guess I don't know what to ask at this point. First, did I miss the
>>> link to the ldom ontology that would clarify the meaning of the ldom
>>> classes and properties here? That would be a start.
>>> 
>>> I still don't see in here a way to address properties or groups of
>>> properties that are not defined as a class. I also cannot understand
>>> if this solution expects that shapes must be coded as such in the
>>> instance data. I can say that I expect to be working with instance
>>> data that is unaware of intended validation, and therefore the nature
>>> of (the shape of) the data itself must be sufficient.
>>> 
>>> Spelling out those aspects would be very helpful. Any single example
>>> is just a single example, and doesn't show the full extent of the
>>> functionality. Examples are helpful to show a single instance of what
>>> you are describing, but they are not in themselves descriptive because
>>> each example shows only one possible case.
>>> 
>>> kc
>>> 
>>>> On 2/11/15 9:43 PM, Holger Knublauch wrote:
>>>> Hi Karen,
>>>> 
>>>> I have added two preamble paragraphs that may help set the stage of what
>>>> is intended.
>>>> 
>>>> https://www.w3.org/2014/data-shapes/wiki/Shape_Selectors
>>>> 
>>>> Please follow up with questions if you have any.
>>>> 
>>>> HTH
>>>> Holger
>>>> 
>>>> 
>>>>> On 2/12/2015 9:42, Karen Coyle wrote:
>>>>> Holger, I find this to be a bit sparse of explanations. I know that
>>>>> it's easier to produce code than lots of verbiage, but I, for one,
>>>>> would appreciate a more ample natural language description of what is
>>>>> intended.
>>>>> 
>>>>> Thanks,
>>>>> kc
>>>>> 
>>>>>> On 2/10/15 3:43 PM, Holger Knublauch wrote:
>>>>>> I have just pushed a new "meta" feature to the LDOM system vocabulary
>>>>>> that can be used to represent which properties shall be used to drive
>>>>>> the LDOM engine. This is an attempt to formalize and generalize the
>>>>>> issue of classes-vs-shapes into a solution that everyone could live
>>>>>> with.
>>>>>> 
>>>>>> Here are the details:
>>>>>> 
>>>>>> https://www.w3.org/2014/data-shapes/wiki/Shape_Selectors
>>>>>> 
>>>>>> I believe this offers the maximum flexibility so that platforms can
>>>>>> pick
>>>>>> which properties (such as oslc:instanceShape/oslc:classShape) they
>>>>>> would
>>>>>> like to use. This approach allows applications to pick whether they
>>>>>> want
>>>>>> backward-compatibility with RDFS classes or use stand-alone shapes, or
>>>>>> even both!
>>>>>> 
>>>>>> Any feedback?
>>>>>> 
>>>>>> Thanks,
>>>>>> Holger
> 
> -- 
> Karen Coyle
> kcoyle@kcoyle.net http://kcoyle.net
> m: 1-510-435-8234
> skype: kcoylenet/+1-510-984-3600
>

Received on Friday, 13 February 2015 15:11:16 UTC