Re: Problems with unrestricted usage of syntax in option 3

Thanks Enrico for responding to my question during the call yesterday.

TL/DR: I still don't get what you find problematic in the examples that 
you give. More detail below

On 16/02/2024 17:59, Franconi Enrico wrote:
> Problems arising with unrestricted usage of syntax in option 3, due to 
> an implicit meaning given to /triple terms/ and to the /rdf:nameof/ 
> property:
>
> Implicit equalities in RDF entailment:
>
> <( :s1 :p1  "42"^^xsd:integer )> :p :o .
> /*entails and is entailed by*/
> <( :s1 :p1 "042"^^xsd:integer )> :p :o .
I agree -- in D-entailment where D contains xsd:integer,
for which we already have:

"42"^^xsd:integer :p :o .
/*entails and is entailed by*/
"042"^^xsd:integer :p :o .

So, how is it making D-entailment more complex than it already is?

Also, even if we restrict triple terms to appear only as the object of 
rdf:nameOf,
what makes the following inference /less/ problematic than the one above??

:e rdf:nameOf <( :s1 :p1  "42"^^xsd:integer )>.
/*entails and is entailed by*/
:e rdf:nameOf <( :s1 :p1 "042"^^xsd:integer )>.

>
> <( :s1 :p1 :o1 )> :p :o .
> <( :s2 :p1 :o1 )> owl:same-as _:b .
> :s1 owl:same-as :s2 .
> /*entails*/
> _:b :p :o .

In OWL entailment, yes. Is it significantly harder to deal with the 
following?

:s1 :p :o .
:s2 owl:same-as _:b .
:s1 owl:same-as :s2 .
/*entails*/
_:b :p :o .

To be clear, the concern that I expressed yesterday was of a different 
nature:
/simple entailment/, currently, provides no way to express equality.
and enforcing rdf:nameOf to be functional /in all simple interpretations/,
we introduce a way to express equality.
This is the  line that I would avoid to cross.

I'm less concerned if we introduce this in a /higher/ form of 
entailment, that already has a notion of equality built-in.

>
> *What is the meaning of:*
> _:b rdf:nameof _:b .
> <( :s1 :p1 :o1 )> rdf:nameof "42"^^xsd:integer .
Is it more problematic than the meaning of the graph below?

_:b rdf:type "foobar".
_:b rdf:first 1, 2, 3.

Although the graph above would be considered strange or even wrong by 
most people, it is syntactically valid, and consistent under simple 
entailment.

>
> I suspect that with a well-formedness condition such as "tripleTerms 
> can only appear as objects of /rdf:nameOf/ triples, which can have 
> only tripleTerms as objects" we could avoid these problems.
> I strongly believe also that nobody is ever interested to write 
> non-well-formed RDF graphs.
Agreed, but I'm not sure that everybody is ready to pay the 
computational cost of enforcing well-formedness either -- again, for 
simple-entailment and SPARQL processing.
> But if we do assume this condition, then why not having natively 
> triple occurrences (i.e., NOT as a macro) and avoid completely the 
> rdf:nameOf property:
> << :wed-1 | :liz :spouse :richard >>
> :starts 1964 ;
> :ends 1974 .
I responded to this in the other thread: 
https://lists.w3.org/Archives/Public/public-rdf-star-wg/2024Feb/0027.html

>
> —e.
>
>