Re: Proposal for RDF-star Minimal Baseline - Agenda item for this week's plenary

Dear Peter,

I am aware of the consequences of making RE injective, I just really do not see why we need that. So, to answer directly:

> Am 18.07.2024 um 20:34 schrieb Peter F. Patel-Schneider <pfpschneider@gmail.com>:
> 
> Suppose we don't include the injection and we have (in an extension of RDF with equality and inequality)
> 
> :a /= :a1 .
> <:a :b :c> = <:a1 :b :c> .
> 
> i.e., <:a :b :c> and <:a1 :b :c> have different components but nonetheless are the same triple.
> 
> Does this ever make sense?   If not, then why not include the injection?

For me, that does make sense. We talk about a triple term, it could mean anything just like an iri. 
But I really like that you share your point of view and your expectations here. So for you, that it would be desirable to not allow the above?

> 
> 
> Suppose we have (in RDFS++)
> 
> <:a :b :c> owl:sameAs <:a1 :b1 :c1> .
> 
> Should this entail
> 
> :a owl:sameAs :a1 .
> :b owl:sameAs :b1 .
> :c owl:sameAs :c1 .
> 
> That is, the same triple has the same components.

No, in my opinion, it should not entail that these things are equal. I even consider that counter-intuitive to get these entailments. But I am willing to adopt if that is what we want and need as a group. I really like that we talk about expectations we have. I think we should even collect such expectations, to be able to test our final proposal on these.

> 
> If so, then you have the injection.

I know, but for me that is an argument against the injection. 


Thank you for your answer.

Kind regards,
Dörthe


> 
> peter
>

Received on Thursday, 18 July 2024 22:20:47 UTC