Re: What are the differences between a quoted triple and its singleton set?

The charter of the RDF-star working group is quite constrained and appears to 
me to not include anything about augmenting the semantics of named graphs or 
making any connection between quoted triples and named graphs.  If you think 
that this was a mistake feel free to get W3C to modify the charter of this 
working group.

peter


On 6/12/23 13:01, Thomas Lörtsch wrote:
> 
> 
>> On 27. May 2023, at 14:45, Peter F. Patel-Schneider <pfpschneider@gmail.com> wrote:
>>
>> Although there is a FPWD of RDF 1.2 Semantics, it does not contain any semantics for quoted triples as the semantics for quoted triples has not yet been determined by the RDF-star working group.
>>
>> So the wording in RDF Semantics has not yet been adjusted to consider unasserted triples.  My view is that this the wording will be adjusted to cover only asserted triples, making your question moot.
> 
> I don’t understand the question as to be only about quoted embedded triples (i.e. embedded RDF-Star triples with a semantics as proposed by the CG report). Any kind of semantics for embedded triples will not answer the question how embedded (quoted or not) triples relate to singleton named graphs (for which the same semantics, quoted or not, could be defined).
> 
> IMO going this route would need
> - a means to express a certain semantics for a named graph (not hard to do, the RDF 1.1 WG just didn’t define it because of lack of prior implementation experience)
> - a syntax for nested triples (if we can update the specifcations for << … >> embedded triples we could just as well update them for {{{ … }}} nested graphs (with and without names)
> 
> There sure are devils in the details, but there always are. This approach would most probably be more elegant and streamlined than the unexplained duality of named graphs and embedded triples. Olaf has called them "orthogonal", Andy upholds that named graphs can only be used for out-of-band administrative purposes. I disagree with both.
> 
> Also implementors disagree, see Amazon Neptune and MilleniumDB where neither statement identifiers nor named graphs are necessarily hardwired into the code, but rather one can substitute the other.
> 
> Anyway, the CG largely ignored the issue, but the WG will have to discuss it. See https://github.com/w3c/rdf-concepts/issues/46.
> 
> Thomas
> 
> 
>> (Note that a quoted triple is a triple occurring as the subject or object of another triple and may also appear as an asserted triple.   Careful distinctions may have to be made in RDF Semantics to keep everything straight.)
>>
>> peter
>>
>>
>>
>>
>> On 5/27/23 08:34, Niklas Lindström wrote:
>>> Dear all,
>>> Given that RDF Semantics (both in 1.1 [1] and 1.2 [2]) state:
>>>      A triple is identified with the singleton set containing it, so
>>> that each triple in a graph is considered to be a subgraph.
>>> Why, then, isn't a quoted triple, say:
>>>      prefix dc: <http://purl.org/dc/terms/>
>>>      prefix : <http://example.org/ns/>
>>>      << <http://example.net/s> :p "o" >> dc:date "2023" .
>>> simply identical to:
>>>      <data:application/n-triples,:%3Chttp%3A//example.net/s%3E%20%3Chttp%3A//example.org/ns/p%3E%20%22o%22%20.>
>>> dc:date "2023" .
>>>      graph <data:application/n-triples,:%3Chttp%3A//example.net/s%3E%20%3Chttp%3A//example.org/ns/p%3E%20%22o%22%20.>
>>> {
>>>          <http://example.net/s> :p "o" .
>>>      }
>>> The one difference I can make is that it is not *explicit* that the
>>> singleton set above is not asserted. Rather, due to [3], it appears
>>> undefined, or user/application dependent(?). That might be amended by
>>> using reification, like:
>>>      <data:application/n-triples,:%3Chttp%3A//example.net/s%3E%20%3Chttp%3A//example.org/ns/p%3E%20%22o%22%20.>
>>> dct:date "2023" .
>>>      <data:application/n-triples,:%3Chttp%3A//example.net/s%3E%20%3Chttp%3A//example.org/ns/p%3E%20%22o%22%20.>
>>> {
>>>          <data:application/n-triples,:%3Chttp%3A//example.net/s%3E%20%3Chttp%3A//example.org/ns/p%3E%20%22o%22%20.>
>>> a rdf:Statement ;
>>>              rdf:subject <s> ;
>>>              rdf:predicate :p ;
>>>              rdf:object "o" .
>>>      }
>>> Or, perhaps better, it could be solved by defining a symbol for
>>> stating that a particular graph is not asserted, e.g.
>>> `blogic:NeutralGraph` (inspired by [4] and [5]), like:
>>>      <data:application/n-triples,:%3Chttp%3A//example.net/s%3E%20%3Chttp%3A//example.org/ns/p%3E%20%22o%22%20.>
>>> a blogic:NeutralGraph .
>>> I've tried to follow the history and ongoing work of RDF-star, and
>>> while I suspect the options above have been considered and rejected, I
>>> fail to grasp exactly why. Is it that since *just* singleton graphs
>>> are insufficient for the criteria of quoted triples, elaborating on it
>>> was not considered worthwhile? Or is it because of how quad stores
>>> work in practise? Perhaps, hidden behind this from the community
>>> report [6]:
>>>      RDF-star describes quoted triples, which are not necessarily
>>> present in any named graph, or within the default graph.
>>> lies a set of unresolved practical problems? E.g. that a graph
>>> containing quoted triples would have to relate to lots of singleton
>>> sets (graphs identified by a triple), and that these sets would need
>>> to be treated differently depending on their type and/or
>>> relationships? I may miss something obvious here, because to me,
>>> introducing a new kind of RDF "atom" just sidesteps that, instead of
>>> working to make more sense of the existing practises for named graphs,
>>> and in the process eliminating unnecessary differences (the divergence
>>> of quotation and named graphs), as well as illuminating, existing
>>> necessary ones (such as how named graphs can be used together in
>>> different ways).
>>> Instead, RDF is being extended with the notion of a recursively
>>> defined "complex triple" (and might even have "levels of compliance"
>>> [7] with ensuing new differences [8]). This is quite a departure from
>>> what RDF has been for over 20 years, where hitherto everything has
>>> been possible to reduce to a simple, *flat* substrate of quads. I'm
>>> trying to understand why this is irrevocably necessary.
>>> While I see many use cases for quoted triples (and for quotation in
>>> general), from my point of view they are *closely* related to the
>>> current practises of named graphs. I fail to see harmony with that in
>>> the evolving semantics. I don't see the necessary difference.
>>> Best regards,
>>> Niklas
>>> [1]: https://www.w3.org/TR/rdf11-mt/#notation-and-terminology
>>> [2]: https://w3c.github.io/rdf-semantics/spec/#notation
>>> [3]: https://www.w3.org/TR/rdf11-datasets/#each-named-graph-defines-its-own-context
>>> [4]: https://www.w3.org/2009/07/NamedGraph.html#named-graphs
>>> [5]: https://www.slideshare.net/PatHayes/blogic-iswc-2009-invited-talk
>>> [6]: https://w3c.github.io/rdf-star/cg-spec/2021-12-17.html#trig-star
>>> [7]: https://github.com/w3c/rdf-star-wg/issues/19
>>> [8]: https://github.com/w3c/rdf-star-wg/issues/33
>>
> 

Received on Monday, 12 June 2023 17:52:49 UTC