Re: shapes as classes

No need to pursue this dialog further. It was interesting enough to visit some ideas from the past. I remember liking the ensemble idea back in the early object days. Consider this thread closed.

Ralph
rhodgson@topquadrant.com




On Dec 12, 2014, at 4:23 PM, Peter F. Patel-Schneider <pfpschneider@gmail.com> wrote:

> I find the connection tenous, at best, unless you mean that ensemble classes are something like classes and classes are something like shapes.  I didn't see anything about validating ensemble classes against data or even recognizing ensemble classes in data.
> 
> If all you want is some old reference to classes, then there are very, very many better ones.  You could go back to the 1970s at the very least, with work on semantic nets and frames.
> 
> peter
> 
> 
> On 12/12/2014 12:58 PM, Ralph TQ [Gmail] wrote:
>> The work on ensembles pre-dates RDF/OWL. I am not saying that an ensemble is
>> what a shape is, I am saying that an ensemble type is a type of shape. And you
>> are correct, there was no concept in the work done at this time on formalizing
>> an ensemble type. Note that I prefer to say ‘type’ as opposed to ‘class’,
>> especially when I know there is also the notion of ‘type classes’ in some
>> functional programming languages (:-)).
>> 
>> I referred to ensembles simply because I think that they are related to some
>> of the stories we have.
>> 
>> Ralph
>> rhodgson@topquadrant.com <mailto:rhodgson@topquadrant.com>
>> 
>> 
>> 
>> 
>> On Dec 12, 2014, at 3:41 PM, Peter F. Patel-Schneider <pfpschneider@gmail.com
>> <mailto:pfpschneider@gmail.com>> wrote:
>> 
>>> It seems to me that ensembles are considerably less general than resources
>>> in RDF graphs because of the flat grouping, and ensemble classes are
>>> considerably less general than shapes or constraints.  These ensemble
>>> classes look more like very simple descriptions or classes, and not much
>>> like shapes.  There doesn't seem to be any notion of validating ensembles
>>> against ensemble classes.
>>> 
>>> peter
>>> 
>>> On 12/12/2014 10:59 AM, Ralph TQ [Gmail] wrote:
>>>> I think of some shapes as  “Ensemble Types”. An ensemble was a concept
>>>> introduced some decades ago for object-oriented analysis and design. See this
>>>> paper by Fichman and Kemerler for more  -
>>>> http://csis.pace.edu/~marchese/CS775/Papers/Fichman_oo_conv_analy.pdf
>>>> 
>>>> "De Champeaux’s ensembles and ensemble classes are the most rigorously defined
>>>> of the clustering mechanisms. Ensembles are analogous to conventional objects,
>>>> while ensemble classes are analogous to conventional classes. An ensemble is a
>>>> flat grouping of objects (or other ensembles) that naturally go together -
>>>> usually because they participate in whole-to-part relationships. An
>>>> automobile. for example, is an ensemble consisting of an engine, doors,
>>>> wheels, etc. “
>>>> 
>>>> The reference to De Champeaux is: D. De Champeaux and P. Faure, “A Comparative
>>>> Study of Object-Oriented Analysis Methods.” J. Oriented-Oriented Programming.
>>>> Vol. 5. No. 1. 1992. pp. 21-33.
>>>> 
>>>> Ralph Hodgson, @ralphtq <http://twitter.com/ralphtq>
>>>> 
>>>> TopQuadrant, Inc.,www.topquadrant.com <http://www.topquadrant.com/>
>>>> <http://www.topquadrant.com/>@TopQuadrant <http://twitter.com/topquadrant>
>>>> 
>>>> /cell: +1 781-789-1664 / fax: 703 299-8330 / main: 919 300-7945/
>>>> */Blog:/*/The Semantic Ecosystems Journal
>>>> <http://www.topquadrant.com/the-semantic-ecosystems-journal/>/
>>>> /
>>>> /
>>>> On Dec 12, 2014, at 12:40 PM, Peter F. Patel-Schneider
>>>> <pfpschneider@gmail.com <mailto:pfpschneider@gmail.com>
>>>> <mailto:pfpschneider@gmail.com>> wrote:
>>>> 
>>>>> If "Shape" is more general than "Class", shouldn't all classes be shapes?
>>>>> 
>>>>> peter
>>>>> 
>>>>> 
>>>>> On 12/11/2014 12:40 PM, Holger Knublauch wrote:
>>>>>> I think my write-up makes it pretty clear that shapes are not classes, but
>>>>>> some classes are shapes. "Shape" is more general than "Class".
>>>>>> 
>>>>>> According to your definition in the glossary, a recognition condition defines
>>>>>> a new named term. Shapes can be used anonymously, e.g. as a nested structure
>>>>>> within another shape, and do not necessarily have to produce named terms.
>>>>>> 
>>>>>> I have likely misunderstood your point, but I am at this stage not sure what
>>>>>> issue you have specifically. Also please feel free to edit the wiki page - I
>>>>>> was definitely only creating a starting point and did not intend to speak on
>>>>>> behalf of the whole group.
>>>>>> 
>>>>>> Thanks
>>>>>> Holger
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> On 12/12/14, 4:53 AM, Peter F. Patel-Schneider wrote:
>>>>>>> I am uncomfortable with the group saying that shapes are RDF classes, as in
>>>>>>> 
>>>>>>> https://www.w3.org/2014/data-shapes/wiki/Resource_Shape_Association
>>>>>>> 
>>>>>>> particular in conjunction with anything that implying that shapes provide
>>>>>>> recognition conditions.  I think that this would put forward the notion that
>>>>>>> the working group is advocating that RDF should be extended to have
>>>>>>> recognition conditions on its classes.
>>>>>>> 
>>>>>>> 
>>>>>>> I say this even though OWL classes to provide recognition conditions, and
>>>>>>> can be considered to be shapes.
>>>>>>> 
>>>>>>> peter
>>