Re: expanding work from quoted triples to graph terms

On 20/10/2023 08:27, Olaf Hartig wrote:
> Hi Peter,
> 
> Regarding the SPARQL work that needs to be done (to extend SPARQL for
> an RDF with quoted triples / triple terms), I would say that the main
> conceptual work is already done (available in the CG report) and the
> only thing that remains is to carry these extensions over into the
> corresponding REC-track documents.
> 
> Why did this not happen so far? As one of the editors of the main
> SPARQL-related document (sparql-query), and probably the same holds for
> the other editors, I was hesitant to start putting any effort into this
> while it appeared that RDF was still a moving target. More
> specifically, it was not clear whether a new definition of the
> semantics of quoted triples would require a change to the evaluation
> semantics of BGPs (as currently defined in the CG report). Recently, as
> far as I understand because I wasn't there, one of the Semantics Task
> Force meetings confirmed that such a change would not be necessary.

If quoted triples become graph terms, the same argument applies. A BGP 
is matched by seeing which setting of variables is entailed by the data 
graph. Implementation-wise it would be similar functionally - but of 
course optimization may be found to do better.

Some of the proposals talk about multiple occurrences. I'm not clear 
whether they are proposing that occurrences go directly into the RDF 
data model or whether the proposal has syntax to related to occurrences. 
If they are in the data model, the argument probably doesn't hold any 
more due to cardinality.

> While this would have been the point to start carrying over the SPARQL-
> related content from the CG report, roughly at the same time, the
> discussion about moving to graphs rather than quoted triples emerged,
> which again made me wait with putting any effort into carrying over the
> SPARQL-related content.
> 
> Yet, again, if the group decides to stick to quoted triples, it will
> not be difficult to get done with the related changes for SPARQL.
> 
> Best,
> Olaf
> 
> 
> On Thu, 2023-10-19 at 12:56 -0400, Peter F. Patel-Schneider wrote:
>> The charter of this working group is very explicit its goal is to add
>> quoted
>> triples to RDF, and no other additions.  Renaming this new construct
>> to triple
>> terms is a reasonable change as all it does is change terminology and
>> not
>> implementation.  But my view is that expanding from quoted triples to
>> quoted
>> graphs or graph terms is a major expansion.  I'm not against this
>> expansion
>> but I believe that the working group should first do the work that it
>> is
>> chartered to do and only then explore something else.
>>
>> So what else is needed to add triple terms to the RDF documents
>> (ignoring the
>> SPARQL documents for now)?   As far as I can tell the only technical
>> work for
>> the RDF documents that remains is semantics.   But semantics appears
>> to be
>> dependent of use cases.
>>
>> So my suggestion is that the working group take a look at the use
>> cases (and
>> potentially submit new use cases) and determine which ones the
>> working group
>> will support.  Then a semantics that supports these use cases can be
>> devised.
>> Once this is done then the working group can look at graph terms if
>> it so decides.
>>
>> There is also SPARQL work to be done and this should also be
>> prioritized over
>> work on graph terms.
>>
>> peter
>>

Received on Saturday, 21 October 2023 17:20:30 UTC