Re: Shape membership without ontological commitment (Problem IRI's are global, shapes are local)

On 16/02/15 17:18, Karen Coyle wrote:
> oops. see below.
>
> On 2/16/15 7:15 AM, Jerven Tjalling Bolleman wrote:
>>
>>
>> On 16/02/15 16:03, Karen Coyle wrote:
>>>
>>>
>>> On 2/16/15 2:58 AM, Jerven Tjalling Bolleman wrote:
>>>> Problem: IRI's are global, shapes are local (or should be) as nicely
>>>> illustrated in this example by Karen.
>>>>
>>>> http://lccn.loc.gov/75300479
>>>>     ldom:hasShape ex:bookShape ;
>>>>     dct:title "Moby Dick" ;
>>>>     dct:creator <http://id.loc.gov/authorities/names/n79006936> ;
>>>>     dct:publisher "M. Kennerley" .
>>>>
>>>> I think Karen is right and that this is a "wrong" assertion.
>>>> I believe it should be asserted as something like this.
>>>>
>>>> http://lccn.loc.gov/75300479
>>>>     ldom:hasShape [ a ex:bookShape ;
>>>>                     ldom:context ex:KarensBookShape ;
>>>>                     ldom:dataContext <> ];
>>>>     dct:title "Moby Dick" ;
>>>>     dct:creator <http://id.loc.gov/authorities/names/n79006936> ;
>>>>     dct:publisher "M. Kennerley" .
>>>
>>> No, actually, the original example is correct. My point is that my
>>> desire to validate data does not translate into the data creator's
>>> knowledge of my validation rules. There may be dozens of different
>>> validation rules being applied to the same data by different users.
>>> Think about making use of LOD data from a variety of sources who are
>>> unaware of your use.
>> I see your usecase, but I am surprised to read that you think the
>> earlier version is better, when you are worried about ontological
>> commitment. I am completely confused now.
>
> Sorry. Too much of a hurry. I read the first as being my original
> example, which was:
>
> http://lccn.loc.gov/75300479
>      dct:title "Moby Dick" ;
>      dct:creator <http://id.loc.gov/authorities/names/n79006936> ;
>      dct:publisher "M. Kennerley" .
>
> None of the versions with a validation-specific shape or class in the
> instance data works for me. My use case is in sharing data between
> loosely (or not at all) connected organizations and data stores. We also
> must assume that our data, which will be open to the world, will be
> re-used in ways we cannot anticipate and perhaps by folks with whom we
> have no communication.
>
>>
>> The first form states that  http://lccn.loc.gov/75300479 has a shape and
>> this is always true. The second says it has a shape but is only true in
>> a certain context.
>
> Right you are. Of the forms of shapes embedded in instance data, this
> one is preferable.
OK, good I still think it can use some tuning to get it "right" but good 
to know I am thinking towards the right direction for your usecase.
>
>>>
>>> I already have many linked data sources that need to be combined for
>>> some of my functions. I am not operating within only my own enterprise
>>> data. Some of this is data like above; some of it is SKOS data. There
>>> are vocabularies and re-usable graphs that are provided globally and
>>> used by hundreds of different systems.
>> Yes, which is why you want to have many different validation statuses,
>> right? As the validation status is independent of the conceptual data,
>> validation status is a local truth, not a global truth.
>>
>> e.g. Everywhere I am older than 18, but not everywhere it is legal for
>> me to drink in a bar. The fact that I am older than 18 is a global
>> truth. If a bar is allowed to serve me ethanol is a local constraint.
>>>
>>> And, precisely, shapes must be inferred in some cases.
>> And sometimes they must be asserted? or am I reading it to shortly.
>
> Sometimes they CAN or WILL be asserted, but not always. My data makes
> use of data from many different sources. In the example above, my Person
> data comes from the Library of Congress. I have no say over how the LoC
> codes its data. Some other data may come from Wikipedia. Eventually the
> publisher data may be linked off of the ISBN database. None of these
> will carry any information about MY validation rules.
Completely agree! Your usecase is similar to my ones when we do data 
integration on demand.
>
> kc
>
>
>>>
>>> kc
>>>
>>> kc
>>>
>>>
>>>>
>>>> But then it could be this as well
>>>>
>>>> http://lccn.loc.gov/75300479
>>>>     a [ rdfs:subClassOf ex:bookShape ;
>>>>         #subclass or type is both ok for me
>>>>         ldom:context ex:KarensBookShape ;
>>>>         ldom:dataContext <> ];
>>>>     dct:title "Moby Dick" ;
>>>>
>>>> i.e. the choice for using the shape or class word does not make a
>>>> difference. Also in both cases these would be classes per Peter's
>>>> definition. As shape/class membership is asserted and not inferred.
>>>>
>>>> Side effects of asserting a ldom:dataContext is that such a context
>>>> description can then have a cryptographic signature. e.g.
>>>>
>>>>
>>>> http://lccn.loc.gov/75300479
>>>>     ldom:hasShape [ a ex:bookShape ;
>>>>                     ldom:context ex:KarensBookShape ;
>>>>                     ldom:dataContext [ ex:tripleSHA512
>>>> "259ab4abea80d95150....066f331334729ba2" ;
>>>> ex:pgpsignature ex:signer <mailto:example@example.org>
>>>>                  ec:signature  "259ab4abea80d95150....066f331334729ba2"
>>>> ] ];
>>>>
>>>> And the shape membership can then also be signed.
>>>> In other words, introducing a level of indirection between a resource
>>>> and a "shape" allows lots of nice features and removes a lot of
>>>> objections.
>>>>
>>>> Regards,
>>>> Jerven
>>>> On 16/02/15 11:00, Dimitris Kontokostas wrote:
>>>>>
>>>>>
>>>>> On Sat, Feb 14, 2015 at 8:01 PM, Karen Coyle <kcoyle@kcoyle.net
>>>>> <mailto:kcoyle@kcoyle.net>> wrote:
>>>>>
>>>>>
>>>>>
>>>>>     On 2/14/15 9:37 AM, Peter F. Patel-Schneider wrote:
>>>>>
>>>>>             The question is only sensible if one already assumes RDFS
>>>>> or OWL
>>>>>             >semantics. Outside of RDFS/OWL semantics, resources can
>>>>> do all these
>>>>>             >things without being classes.
>>>>>
>>>>>         OK, so replace RDF classes with "RDF types in RDF and RDFS
>>>>>         classes in
>>>>>         RDFS".  The point is whether documents will contain triples
>>>>> that
>>>>>         use shapes
>>>>>         where there are now RDF type or RDFS classes.
>>>>>
>>>>>
>>>>>
>>>>>     If instance data will use shapes where they now use RDF, how would
>>>>>     you fulfill the requirements implied in User Story #4 [1] where
>>>>> the
>>>>>     same node in a graph can serve multiple roles? Or in general
>>>>> how do
>>>>>     you address re-usability of your data in different contexts?
>>>>>
>>>>>
>>>>> Shapes does not solve this problem, maybe postpone it a bit [1] until
>>>>> when/if they get further adopted. In the same way one can use
>>>>> different
>>>>> "shapes" at different contexts, one can use different class
>>>>> constraints
>>>>> at different contexts.
>>>>>
>>>>> [1]
>>>>> https://lists.w3.org/Archives/Public/public-data-shapes-wg/2015Jan/0198.html
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>     kc
>>>>>     [1]
>>>>>
>>>>> https://www.w3.org/2014/data-__shapes/wiki/User_Stories#S4:___Issue_repository
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> <https://www.w3.org/2014/data-shapes/wiki/User_Stories#S4:_Issue_repository>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>     --
>>>>>     Karen Coyle
>>>>>     kcoyle@kcoyle.net <mailto:kcoyle@kcoyle.net> http://kcoyle.net
>>>>>     m: 1-510-435-8234
>>>>>     skype: kcoylenet/+1-510-984-3600 <tel:%2B1-510-984-3600>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> Dimitris Kontokostas
>>>>> Department of Computer Science, University of Leipzig
>>>>> Research Group: http://aksw.org
>>>>> Homepage:http://aksw.org/DimitrisKontokostas
>>>>
>>>
>>
>

-- 
-------------------------------------------------------------------
Jerven Bolleman                        Jerven.Bolleman@isb-sib.ch
SIB Swiss Institute of Bioinformatics  Tel: +41 (0)22 379 58 85
CMU, rue Michel Servet 1               Fax: +41 (0)22 379 58 58
1211 Geneve 4,
Switzerland     www.isb-sib.ch - www.uniprot.org
Follow us at https://twitter.com/#!/uniprot
-------------------------------------------------------------------

Received on Monday, 16 February 2015 16:38:42 UTC