Re: Against the notion of reification well-formed graph (i.e., atomicity)

> On 23. Jan 2024, at 12:22, Peter F. Patel-Schneider <pfpschneider@gmail.com> wrote:
> 
> 
> 
> On 1/23/24 03:24, Franconi Enrico wrote:
>>> On 22 Jan 2024, at 20:46, Peter F. Patel-Schneider <pfpschneider@gmail.com> wrote:
>>> 
>>> The mapping from names to triples is not an injection,
>> From what I read in your document:
>>    Definition: An RDF graph is reification well-formed iff any node in it
>>    that is the subject of a triple in the graph with predicate |rdf:subject|,
>>    |rdf:predicate|, or |rdf:object| must be the subject of *_exactly_* one
>>    triple in the graph with each of these predicates. /[bold/underline is mine]/
>> So, the definition I am reading from you requires a bijection.
>> If I am wrong, then I guess we agree
> 
> This says nothing about names, only rdf:subject, rdf:predicate, and rdf:object links.  It doesn't require *any* bijections.
> 
>>> All well-formedness says is that you can't mess up the reification structures, i.e., you can't add something like
>>> 
>>> :e rdf:subject :t .
>>> 
>>> to the above graph and retain well-formedness.
>> I believe that here we are confusing two distinct aspects:
>> 1. avoiding dangling links in the expansion to :e rdf:subject :s.
>    :e
>>    rdf:predicate :p.
>    :e rdf:predicate :o.
>> 2. having a bijection between triples and their names,
>>    which, in turn, should be clarified whether it is intended
>>       2a.  syntactically or
>>       2b.  semantically
> 
> There is no hint of a bijection anywhere, aside from the requirement that non-literal nodes in an RDF graph have to be either IRIs or blank nodes.  Even this isn't a bijection, really.
> 
>> I believe that (1) should be always enforced, while we should not enforce (2) in neither forms 2.a or 2.b.
> 
> The proposal's notion of RDF graphs is taken directly from /https://www.w3.org/TR/rdf11-concepts/#section-rdf-graph.  Its notion of (simple) semantics is taken directly from https://www.w3.org/TR/rdf11-mt/#simple-interpretations
> 
> In the proposal triples don't have names.  They don't even exist outside of being elements of an RDF graph.  There is no suggestion in the proposal of a bijection between triples and anything.
> 
> What the proposal does talk about is RDF reifications, nodes in an RDF graph that are subjects of rdf:subject, rdf:predicate, or rdf:object triples.  The well-formedness requirement states that an RDF graph is ill-formed if it has a node that is the subject of a triple with any of these predicates and is not the subject of exactly

Shouldn’t this be changed to *at least*? See my prior mail in response to Dörthe.

> one triple with each of these predicates.  No bijection between triples and anything is either mentioned or implied.  The notion of well-formedness is completely syntactic.
> 
>> We should describe in the best practices section the case where users should use RDF* to encode reification with bijection, by just self-imposing such restriction.
>> But there are many cases where this shouldn’t be the case.
>> It seems that we agree here, so I am happy - but please fix the definition.
>> cheers
>> —e.
>>> On 1/22/24 10:51, Franconi Enrico wrote:
>>>> In this message I want to argue against the notion of reification well-formed graph (i.e., atomicity).
>>>> Adding atomicity violates a necessary aspect of the proposal (as written by pfps):
>>>> "There is no intended meaning of the new concrete syntax beyond what it expands to".
>>>> According to atomicity:
>>>> (1)
>>>> <<:a | :s1 :p1 :o1>> :s :o .
>>>> <<:b | :s2 :p2 :o2>> :s :o .
>>>> :s1 :same-as :s2 .
>>>> :p1 :same-as :p2 .
>>>> :o1 :same-as :o2 .
>>>> should entail
>>>> :a :same-as :b .
>>>> Atomicity adds an intended meaning: it assumes that the same triple has systematically the same identifier, and therefore (1) should be a valid entailment.
>>>> If you don’t like entailment (1), then you contradict the intention of atomicity, UNLESS you assume full opacity.
>>>> (2)
>>>> <<:a | :s1 :p1 :o1>> :s :o .
>>>> <<:b | :s2 :p2 :o2>> :s :o .
>>>> :a :same-as :b .
>>>> should entail
>>>> :s1 :same-as :s2 .
>>>> :p1 :same-as :p2 .
>>>> :o1 :same-as :o2 .
>>>> Atomicity adds an intended meaning: it assumes that a resource identifies at most one triple.
>>>> If you don’t like entailment (2), then you contradict the intention of atomicity, UNLESS you assume full opacity.
>>>> So, (1) and (2) show that _EITHER_ you assume full opacity (and therefore you are assuming an intended meaning), _OR_ you accept the above entailments (and therefore you are assuming an intended meaning).
>>>> Note also that entailments like the above make the life of SPARQL BGP matching more complex.
>>>> I also understand that most (if not all) use cases assume transparency.
>>>> (3)
>>>> Atomicity would disallow writing typical reification statements like:
>>>> << :w1 | :bill-clinton :related-to :hillary-rodham >> :starts 1975 .
>>>> << :w1 | :42nd-potus :husband :1st-female-NY-senator >> :starts 1975 .
>>>> cheers
>>>> —e.
>>> 
>

Received on Tuesday, 23 January 2024 11:30:37 UTC