Re: Atomic reification

On 26/01/2024 10:38, Thomas Lörtsch wrote:
> 
> 
>> On 25. Jan 2024, at 18:16, Andy Seaborne <andy@apache.org> wrote:
>>
>> Points discussed about N-triples in the last two meetings e.g. [1], [2] meant I understood "atomic reification" differently.
>>
>> https://github.com/afs/rdf-star-notes/blob/main/reif-atoms.md
>>
>> "well-formed" gives some grouping but that is lost in N-triples. The test for it is expensive [3].
>>
>>
>> To flesh out the problems with RDF 1.1 reification:
>>
>> 1. Visually verbose (Turtle).
>> 2. Verbose number of triples (N-Triples).
>> 3. The reification triples may be spread across parts of a document.
>>    This breaks visual proximity as written.
>> 4. Over-specificed reifications (e.g. multiple rdf:subject)
>> 5. Under-specificed reifications (e.g. no rdf:subject)
>> 6. Presence or absence of rdf:Statement is unclear leading to
>>    different SPARQL results. (De facto, it is omitted.)
>> 7. If a large graph is split into multiple files of manageable size,
>>    the triples can be split across those files (and blank nodes for
>>    reification are broken).

It was clear last week that the original 4 point list was being read in 
different ways by different people. I hope this more detailed list helps.

The atomic reification proposal builds on the agreed Turtle syntax.

"Seeking consensus on RDF-star" section "What we seem to agree on"

https://lists.w3.org/Archives/Public/public-rdf-star-wg/2024Jan/0095.html

> As already mentioned yesterday in a mail [4] to this list I’m not so convinced by this  list.

Some of them may not be of concern to you. No problem.
RDF has diverse interests and we are a diverse WG.

> To cite [4]:
> 
> You give a list of 7 problems with RDF reification.

> Some of them (problems 4, 5 and 6) would be handled by a notion of wellformedness.

Yes.

The wellformedness condition (better seen in the "syntactic sugar 
proposal) is a whole-graph condition that is not maintained on merge.

https://lists.w3.org/Archives/Public/public-rdf-star-wg/2024Jan/0104.html

That's not an absolute reason not to base work on it; it is a factor 
that I think needs to be considered.

"Atomic reification" attempts to make the condition part of the abstract 
data model.

> Problem 1 would be solved by the proposed annotation syntax.

Problem 1 is "Visually verbose (Turtle)".

The doc (AR proposal) says it is solved by the agreed Turtle syntax that 
includes annotation syntax and more.

> Problems 3 and 7, especially blank nodes split over multiple graphs when breaking a big graph into files of a more manageable size, are not specific to reification but a general problem.

point 3 : "reification triples may be spread across parts of a document"
I agree this addressed by Turtle syntax.

It does not apply to N-Triples. N-triples is often machine generated - 
the triples can be anywhere in the document. Because the WG is 
discussing expectation on implementation, I think it is a relevant issue.

point 7 : "If a large graph is split into multiple files of manageable size"

Yes, this is a general problem. RDF Reification is affected by it.
The "wellformedness" condition can be split across the files.

> That leaves problem 2, verbosity in N-Triples, and that just comes with the terrain. There sometimes are more or less verbose ways to represent a complex type in straight triples - RDF Collections are much worse than RDF Containers - but RDF standard reification is not too bad in that respect.
> You mention a reification atom <<(s p o)>> as a possible addition to N-Triples. That seems like a slight variation of N-Triples-Star to me, and I’m not fundamentally opposed if it helps and doesn’t rely on a new term type. 

As an element of the RDF abstract data model - yes, it is a variation of 
N-Triples-Star. Unlike N-Triples-Star it is only in the object position.

The proposal is based around "named occurrences" being in IR.

Complete proposals has been asked for, including semantics, so I had a 
go at doing that (not my natural area).

> My question is: does it really help? And would you also add list atoms, CBD atoms, graph atoms?
> We discussed this already in follow-up mails [5] and [6].
> In a further mail [7] I questioned the relevance of the problem 2 in streaming contexts.

Understanding the impact of WG decisions is important.
No one is saying it is a "decisive" [7] point.

    Andy

> Best,
> Thomas
> 
> 
> [4] https://lists.w3.org/Archives/Public/public-rdf-star-wg/2024Jan/0159.html
> [5] https://lists.w3.org/Archives/Public/public-rdf-star-wg/2024Jan/0160.html
> [6] https://lists.w3.org/Archives/Public/public-rdf-star-wg/2024Jan/0161.html
> [7] https://lists.w3.org/Archives/Public/public-rdf-star-wg/2024Jan/0163.html
> 
>> The agreed Turtle syntax hides some of these issues (it addresses 1, it tends away from 4 and 5). It does not apply for N-triples, and not for SPARQL results when thought of as querying a triple table.
>>
>>
>> The general idea, and as mentioned last week, is to introduce a unit "reification atom" that means the equivalents of rdf:subject/rdf:predicate/rdf:object are not split.
>>
>> The agreed Turtle syntax applies.
>>
>>
>> There has been requests for formal semantics in proposals so I have tried to provide that - I hope people here can help improve it in both presentation and content.
>>
>> https://github.com/afs/rdf-star-notes/blob/main/reif-atoms.md
>>
>> https://github.com/afs/rdf-star-notes/blob/main/reif-atoms-interpret.md
>>
>>     Andy
>>
>> [1] https://www.w3.org/2024/01/18-rdf-star-minutes.html#x123
>> [2] https://www.w3.org/2024/01/19-rdf-star-minutes.html#x169
>> [3] https://lists.w3.org/Archives/Public/public-rdf-star-wg/2024Jan/0136.html
>>
>

Received on Friday, 26 January 2024 12:59:41 UTC