Re: RDF-star profile "transparent" [Was: Re: The way forward]

On 2 May 2024, at 15:13, Antoine Zimmermann <antoine.zimmermann@emse.fr> wrote:
> 
> Le 02/05/2024 à 15:08, Franconi Enrico a écrit :
>>> On 2 May 2024, at 14:58, Antoine Zimmermann <antoine.zimmermann@emse.fr> wrote:
>>> 
>>> Le 01/05/2024 à 17:27, Franconi Enrico a écrit :
>>>> I have written down the formal definition of two profiles in the wiki:
>>>>  * RDF-star profile “transparent”
>>>>    <https://github.com/w3c/rdf-star-wg/wiki/RDF-star-profile-%22transparent%22> (namely many-to-many transparent)
>>> 
>>> This is only about the first profile you propose.
>>> I have a problem with the way triple terms are interpreted.
>>> In your semantics, there is a mapping RE from IR x IP x IR into IR, called the denotation of triple terms.
>>> This is simply saying that, for any sequence of 3 resources, the second of which is a property, there is a corresponding resource in the universe. It does not imply in any way that the first element of the sequence plays the role of a subject, that the third element plays the role of an object, and that the 2nd element is assumed to be relating the other two. As far as this formal semantics is concerned, this 3-uple of resources could simply be 3 unrelated resources of interest that are ordered like this. This is quite different from "being an RDF triple" or being an expression of a statement.
>>> 
>>> This proposal is in fact very similar to the one I proposed a while ago:
>>> 
>>> https://www.emse.fr/~zimmermann/W3C/RDF-star-semantics/

>>> 
>>> except that, additionally, in my proposal, the resource associated with a triple term would have to be connected to the first, second, and third element of the triple term by way of 3 properties. This ensures that when one is interpreting a triple term, it is interpreted with an assumption of what roles the 3 components of the triple term are playing.
>> Those 3 properties you introduce would be redundant, in the same way it would be redundant to require to say /_explicitly_/ that a standard RDF triple s p o. has s as subject, p as predicate, and o as object, while their role is implicit in the ordering.
>> Wouldn’t you agree?
> 
> (…) an RDF triple in a graph gives a special role to each of the three elements, because the middle element (i.e. the predicate) is interpreted as something that has an extension (IEXT in the semantics) that have pairs of entities, and the subject's entity, followed by the object's entity, must be among these pairs. You could not interpret a triple (s, p, o) as a first order logic atom s(p,o), for instance, because of that.

I agree.

> With your semantics, there isn't such structure in the interpretation that gives any particular role to the 3 components, so it's just an arbitrary list of 3 things.

No. The structure is given by the order. Using 3 properties is isomorphic to the order, since each one of the 3 properties corresponds to a position in the triple term.

—e.

>>>>  * RDF-star profile "functional opaque”
>>>>    <https://github.com/w3c/rdf-star-wg/wiki/RDF-star-profile-%22functional-opaque%22> (namely many-to-one opaque)
>>>> They rely on two distinct properties - rdf:reifies and rdf:edge - and on two distinct syntactic categories - tripleTerm and opaqueTripleTerm.
>>>> For this reason, they could be just merged into a unique profile, which actually could be RDF-star itself.
>>>> Let me know comments,
>>>> cheers
>>>> —e.
>>>>> On 25 Apr 2024, at 02:37, Lassila, Ora <ora@amazon.com> wrote:
>>>>> 
>>>>> [My apologies that this comes at the last moment before tomorrow’s meeting.]
>>>>> We have had long discussions within the Neptune team about the ongoing debate in the WG. We want to find an amicable, consensus-based way forward. Obviously the support within the WG for the multi-triple reifier proposal is strong, and we understand that many WG members may not be willing to live with the single-triple reifier approach. That said, we also believe that we (Neptune and our OneGraph project) need to be true to our vision of the future of “graph interoperability”.
>>>>> Thus, we would like to bring back the idea of profiles: one for the multi-triple reifier support, another for the single-triple option. This would allow implementors some leeway, and would ultimately let the graph marketplace choose. People already make choices about what technologies they use, sometimes based on the level of support different technology vendors offer. Bottom line: we do not want to block progress in the WG, and this would let us move towards finishing the specifications. I think it is better that we get the largest possible number of implementors building RDF 1.2 -compliant products, rather than some companies “opting out”.
>>>>> Ora
>>>>> -- 
>>>>> Dr. Ora Lassila
>>>>> Principal Technologist, Amazon Neptune
>>> 
>>> -- 
>>> Antoine Zimmermann
>>> École des Mines de Saint-Étienne
>>> 158 cours Fauriel
>>> CS 62362
>>> 42023 Saint-Étienne Cedex 2
>>> France
>>> Tél:+33(0)4 77 49 97 02
>>> https://www.emse.fr/~zimmermann/

>>> 
> 
> -- 
> Antoine Zimmermann
> École des Mines de Saint-Étienne
> 158 cours Fauriel
> CS 62362
> 42023 Saint-Étienne Cedex 2
> France
> Tél:+33(0)4 77 49 97 02
> https://www.emse.fr/~zimmermann/

>

Received on Thursday, 2 May 2024 13:20:39 UTC