Re: using Shape Expressions to validate RDF graphs

I currently do validation of the sort that I think Peter is talking about, using custom sparql queries, where I verify what roughly corresponds to owl Restrictions (cardinality, allValues, someValues) except at the syntactic level (are there triples present) rather than the semantic level (like a DL reasoner would do)

My understanding of Peter's point about
<Person> {  a (:Person) , :spouse @<Person>? }

is that what we really want to say, is that a person has at most one spouse (depending on cultural and temporal issues) and that spouse is a person, and may have other properties.
Does the shape expression prohibit more than one spouse? Does the shape expression prohibit having a spouse that is a non-Person?

===


(The last time I was discussing this topic in the Semantic Web lists I believe there was some controversy about gender and same-sex marriage …., I am glad we are pass that stage!)

Jeremy




On Jul 11, 2014, at 9:54 AM, Peter F. Patel-Schneider <pfpschneider@gmail.com> wrote:

> 
> 
> On 07/11/2014 09:31 AM, Jose Emilio Labra Gayo wrote:
>> On Fri, Jul 11, 2014 at 5:35 PM, Peter F. Patel-Schneider
>> <pfpschneider@gmail.com <mailto:pfpschneider@gmail.com>> wrote:
>> 
>>    I've been looking through the Shape Expressions submission.  I'm having a
>>    hard time figuring out how it can be used to validate RDF graphs.
>> 
>>    The kind of validation I'm thinking of is constraining the elements of
>>    various RDF types, perhaps requiring something like
>> 
>> 
>> I will give a first try...
>> 
>>    - the spouse of every person is a person
>> 
>> 
>> This could be declared as:
>> 
>> <Person> {  a (:Person) , :spouse @<Person>? }
>> 
>> meaning that the shape Person has type :Person and can have an optional triple
>> :spouse of shape <Person>
> >
> > You can play with the example here: http://goo.gl/pcL0ix
> 
> Yes, but how does that determine whether the spouse of every person is a person?  All that it is doing is determining which nodes match the shape <Person>.
> 
> 
>>    - the university of every graduate student is a research university
>> 
>> 
>> I declared it as:
>> 
>> <GradStudent> { :university @<ResearchUniversity> }
>> 
>> <ResearchUniversity> { a (:ResearchUniversity) }
>> 
>> You can also play with the example here: http://goo.gl/fkx71w
> 
> Same comment.
> 
>> Taking a look at your examples, I would like to comment that there is a
>> difference between the domain of discourse of RDF Data Shapes and OWL.
> 
> I don't see where OWL has come into the discussion so far.  Everything above was stock RDF.
> 
>> OWL is very well suited to describe concepts like Person, Graduate Student,
>> University, etc. while RDF Data Shapes are more oriented towards RDF data
>> representations, although they can be related, it is not always the case.
>> 
>> For example, the constraints that I would express in RDF Data Shapes would be
>> something like:
>> 
>> "In this RDF graph, the representation of a Person (or the shape of the
>> resource Person) has a property rdf:type with value :Person and can have a
>> property spouse which must have the shape of a Person"
> 
> That seems to be what I want to say.  Let me reiterate:
> 
> I want to be able to determine whether, in a particular RDF graph (including RDFS inferences, by the way) all the elements of the class Person have each of their spouses (if any) also belonging to Person.
> 
>> I mean, OWL and Shape Expressions have different goals...with OWL you model an
>> ontology of concepts, while with Shape Expressions you just describe the
>> shapes of RDF graphs.
> 
> Agreed, but the very first thing that I want to do along these lines is to determine whether the elements of a particular class have a particular shape.  I think that this is one of the most essential tasks that need to be done along these lines - think of it as type checking of the information in the RDF graph.
> 
>> Best regards, Jose Labra
> 
> peter
> 
>

Received on Friday, 11 July 2014 18:19:24 UTC