Re: Statement identifiers

> On 24. Jan 2023, at 16:59, Peter F. Patel-Schneider <pfpschneider@gmail.com> wrote:
> 
> Can you provide a pointer to the discussion around how RDF/XML is being used by Jerven?  

https://w3c.github.io/rdf-star/UCR/rdf-star-ucr.html#uniprot-attributed-evidenced-triples
https://w3c.github.io/rdf-star/Minutes/2020-12-04.html

I tried to get the topic at least mentioned in the CG report, but to no avail [0].

> Also, where do statement identifiers show up in RDF/XML

https://www.w3.org/TR/rdf-syntax-grammar/#section-Syntax-reifying

> and how do they impact the RDF graphs?

They are syntactic sugar. In the abstract graph they create a reification (quadlet). 

> I'm also not aware of where a reification quadlet shows up in RDF.  Can you provide a pointer to a discussion on this?  

That is well-known terminology in semweb circles, see for example https://jena.apache.org/documentation/notes/reification.html. See https://www.w3.org/TR/2014/REC-rdf-schema-20140225/#ch_reificationvocab for a more complete description of the underlying RDF concepts and vocabulary.

> It is unclear what exactly nodes that are instances of rdf:Statement should end up representing.  There is non-normative wording in the RDF 1.1 semantics that they refer to a concrete realization of an RDF triple, but what then is the relationship to actual statements is left open.  

Quite right, and that is indeed a problem. It is however just as well a problem in the mapping that you propose, isn’t it? 

The RDFn approach makes the connection syntactically unambiguous, and IMO that’s a very good thing, since the intuition is often that the reification refers to a specific occurrence. Now we just have to develop a fitting semamtics.

> It is certainly not the case that different nodes in an RDF graph that are instances of rdf:Statement have to refer to different statements, even if they have the same subject, predicate, and object.

Well, there is an example in https://www.w3.org/TR/rdf11-mt/#reification of two reifications describing the same statement but representing different occurrences. What more clarification do you require? Of course anybody is free to declare them as being the owl:sameAs each other - that’s up to users.


To quote from another mail:
>> On 16. Dec 2022, at 18:17, Peter F. Patel-Schneider <pfpschneider@gmail.com> wrote:
>> 
>> The RDF semantics itself does not say anything about whether an instance of rdf:Statement is an occurrence indeed anything else.  There is informative wording about the intent of instances of rdf:Statement, but I don't see any reason to prevent carving out some instances of rdf:Satement and making them act more like abstract triples.

Do you intend to redefine RDF standard reification in a way that lets all reifications refer to a type rather than occurrences? Or introduce subclasses of rdf:Statement that refer to types and occurrences respectively?


Thomas


[0] https://github.com/w3c/rdf-star/pull/171


> peter
> 
> 
> 
> 
> On 1/23/23 19:36, Thomas Lörtsch wrote:
>> Reading the WG meeting minutes I get the idea that there is some reservation w.r.t. to statement identifiers in Souri's RDFn proposal as if they were a totally new and unproven idea. They have been around for decades in RDF/XML. Jerven Bolleman reported to the RDF-star CG that the UniProt project still uses RDF/XML because of the nice syntactic sugar that its statement identifiers provide for RDF standard reification. Standard reification has clear semantics: a reification quadlet creates a new identifier that refers to _some_ occurrence of a statement (without addressing that specific occurrence however). Multiple reification quadlets reify different occurrences, so no bag semantics.
>> 
>> This mapping of statement identifiers to RDF standard reification is for multiple reasons not what I would consider sufficient for bridging the gap between LPG and RDF, but it is in line with what many of you seem to advocate: an approach that is merely able to record some provenance and can’t interfere with other statements, jeopardizing monotonicity. And as it is only and purely syntactic sugar for standard reification (*) I really wonder what there is not to understand or even fear about it.
>> 
>> Best,
>> Thomas
>> 
>> 
>> (*) Well, with one caveat: it does even more than a normal reification quadlet misleadingly suggest that it would address a specific statement occurrence as it is syntactically so tightly connected to such. It shares that problem with the shortcut syntax of RDF-star.

Received on Tuesday, 24 January 2023 22:33:00 UTC