Re: shapes-ISSUE-18 (S35 examples): S35 needs to state what constraints are required

Peter, Thx for the feedback.

"Peter F. Patel-Schneider" <pfpschneider@gmail.com> wrote on 01/08/2015 
08:10:58 PM:

>  > Let X be  the URI of an access control list information resource.
> 
> Information resources can in general be accessed by multiple URIs.  What 

> happens then?

Agreed. This is an informal description. In a formal description I'd say: 

"Let X be an IRI and let G be an RDF graph."

> 
>  > Its RDF graph must contain X as a resource node.
> 
> "Resource node" is not a term defined by RDF.  I think that just 
> "node" is fine.

Constraint 1: G must contain X as a node.

> 
>  > X must have type acc:AccessContextList.
> 
> I assume here that you mean that X must be linked to 
acc:AccessContextList by 
> an rdf:type triple.

Constraint 2: G must contain a triple (X rdf:type acc:AccessContextList)

> 
>  > X must have a string-valued dcterms:title property and a 
string-valued
> dcterms:description property.
> 
> I assume that you mean "... value for ..." instead.

Constraint 3: G must contain a triple that matches (X dcterms:terms ?O) 
where ?O is a literal of type xsd:string

etc. 

> You seem to be implying that the working group should define an API 
> that takes 
> a URI and then somehow coerces this to an RDF graph.  This may or 
> may not be a 
> good idea, but it has nothing to do with disconnected graphs.  If you 
think 
> that this should be a requirement then I suggest changing the title of 
S35.

No. I am saying that the input to the above is a pair (X, G).

> 
> You also seem to be implying that the working group should define a 
mechanism 
> that somehow gives the constraints access to the URI that was used to 
access 
> the RDF graph.   This may or may not be a good idea, but it has nothing 
to do 
> with disconnected graphs.  If you think that this should be a 
> requirement then 
> I suggest changing the title of S35.

I am saying that we need to be able to express constraints on RDF graphs, 
and those constraints may be parameterized. In this case the parameter is 
X.

-- Arthur Ryman

Received on Thursday, 29 January 2015 14:38:13 UTC