Re: Proposed Resolution for Issue 42

On 31 May 2011, at 21:08, Juan Sequeda wrote:
> Thinking quickly out loud... so please correct me if I'm wrong
> 
> wouldn't we need rdfs:range triples for the case of reference triples:
> 
> <Student/id=1> <enrolled> <Course/id=2>

I don't think so. We already have in the direct graph:

   <Course/id=2> a <Course>.

That's sufficient to know that <Course/id=2> is the result of mapping a tuple in the Course table. The fact that <Course/id=2> occurs as the object of <Student#enrolled> is sufficient to infer the rdfs:range statement.

An explicit rdfs:range triple would be more convenient, I don't dispute that :-) But not strictly necessary. So I'd prefer to keep this discussion separate from ISSUE-42.

In terms of process, I think it would be good if someone created an issue for these additional direct mapping schema triples, and put together a quick strawman on the wiki 

Richard


> 
> 
> Juan Sequeda
> +1-575-SEQ-UEDA
> www.juansequeda.com
> 
> 
> On Tue, May 31, 2011 at 2:53 PM, Richard Cyganiak <richard@cyganiak.de> wrote:
> Hi Alexandre,
> 
> On 31 May 2011, at 20:27, Alexandre Bertails wrote:
> >> I would agree to a proposal that maintains reversibility of the mapping by adding rdfs:domain triples to the properties, and does not generate triples for NULL values.
> >
> > I think that for the moment, we can agree on the current proposal
> > without speaking about any concrete solution, which will come later when
> > we're ready for it.
> 
> No, because I'd like to know what I am agreeing to. I would likely be opposed to a solution that introduces a parliament of OWL into the direct mapping in order to work around the NULL issue.
> 
> > rdfs:domain may be enough for this issue, but we may want other information as well.
> 
> I think we all agree that rdfs:domain is *necessary*.
> 
> I believe that it is also *sufficient* to reconstruct the NULLs, and have seen no claims to the contrary.
> 
> So let's go with rdfs:domain *only* as the resolution to ISSUE-42.
> 
> More schema triples may still be added to the direct mapping later on, but that needs to be discussed, and it can't be discussed before there's a proposal on the table. So I suggest treating additional schema triples as a different and separate issue (which someone should create in the tracker).
> 
> PROPOSAL: Resolve ISSUE-42 by not creating triples for NULL values, and adding rdfs:domain statements to the direct mapping graph. This does not preclude adding more schema triples in a future resolution.
> 
> Best,
> Richard
> 

Received on Tuesday, 31 May 2011 20:25:59 UTC