Re: assertions on facts

Good afternoon James,

(The following includes mostly verbatim quotes from the referenced
documents, turned into a numbered list and with some added emphasis.)

In Concepts [1], the notion RDF statement is defined as:

1. *Asserting an RDF triple*, or *saying* that some relationship, indicated
by the predicate, *holds* between the resources denoted by the subject and
object (as explained below, not all triples are asserted).

Such a statement, thus *corresponds* to an RDF triple (it is not denoted by
the triple, it *asserts* it); and by association in an interpretation, it
corresponds to a proposition denoted by a triple. Does its existence *make*
it true? No; you either trust it or you don't (this is not the domain of
formal semantics, it is either the domain of trust and/or the domain of
epistemology; both of whom ideally makes use of the former). You need to
trust the veracity of a document or message having resulted from the act of
making one or more such statements, and somehow determine that trust by
assessing statements; perhaps including statements about statements.

Therefore, this statement notion is not further defined formally; its
meaning pertains to a domain of discourse, which can be described through
RDF vocabulary and corresponding definitions. The rdf:Statement class in
RDF Schema is defined as the class of statements "made by a token of an RDF
triple".

Also in Concepts [2], it is noted that:
2. The subject of a triple with the predicate rdf:reifies is called a
reifier.
3. Reifiers may denote a variety of things that are related to (i.e. reify)
propositions (such as a *statement* or belief that the proposition holds,
or an event or circumstance that makes the proposition true).
4. There can be multiple, distinct reifiers related to the same abstract
proposition, such as *statements with different sources*, or circumstances
with different characteristics.
5. One reifier may also be used to reify multiple, distinct propositions,
expressing different propositional aspects using the same reifier.
6. It is possible to make *statements about any kind of statement*,
including an *unasserted statement* that contradicts another statement,
whether asserted or not.

In Semantics [3], it is established that:
7. The set of facts is the set of propositions which are true in the
interpretation.

With this it is defined how to determine whether, for example, a statement
reifies a fact (by checking if the proposition reified is among the facts).

In my opinion, this more than covers what the charter contains wrt. "the
notion of quoted triple to express statements about statements". (The
charter also allows the WG to "reconsider this and proceed differently from
the Community Group's proposal".)

Best regards,
Niklas

PS. I agree with Peter's reply as well; including that Schema might need
some more care. I raised [4] in part to address that.

[1]: <https://w3c.github.io/rdf-concepts/spec/#dfn-rdf-statement>
[2]: <
https://w3c.github.io/rdf-concepts/spec/#section-triple-terms-reification>
[3]: <https://w3c.github.io/rdf-semantics/spec/#simple_entailment_properties
>
[4]: <https://github.com/w3c/rdf-schema/issues/57>


On Thu, Aug 28, 2025 at 3:22 PM James Anderson <
anderson.james.1955@gmail.com> wrote:

> good afternoon;
>
> > On 22. Aug 2025, at 17:35, Peter F. Patel-Schneider <
> pfpschneider@gmail.com> wrote:
> >
> > In the meeting yesterday there was discussion on whether RDF 1.2
> currently supports assertions on facts (asserted triples).  My claim is
> that it does, and in a way that needs no change to Semantics.
> >
> > During the meeting there was some discussion about some text that I
> authored. I believe that the text is
> > https://github.com/w3c/rdf-concepts/pull/220#issuecomment-3113421890
> >
> > That text, in essence, says that Semantics provides the basic machinery
> to talk about assertions on facts because it defines facts in an
> interpretation. This is then used to define facts asserted by a graph.
> That's all the semantic machinery that is needed to support assertions on
> facts.
> >
> >
> > The remaining part is to call on the intended meaning of rdf:reifies and
> say (not in Semantics) that an assertion about a fact in a graph is a
> triple whose subject is a reifier of a triple where the object of the
> reification triple is (denotes) a fact in the graph.  So ex:r ex:source
> ex:NYT is an assertion about the fact ex:John ex:loves ex:Mary in
> >
> > ex:r rdf:reifies <<( ex:John ex:loves ex:Mary )>> .
> > ex:John ex:loves ex:Mary .
> >
> > This remaining part is not suitable for Semantics, in my opinion, as it
> does not provide any semantic machinery but is instead about the intended
> meaning of rdf:reifies and how rdf:reifies is supposed to be used.  As far
> as Semantics is concerned, any property could be used.  This is a decided
> positive, as it allows any property to be used in a similar way, perhaps
> with a stronger intended meaning or perhaps with a weaker intended meaning.
>
>
> in the rdf 1.2 semantics document, rdf:reifies appears in four different
> term lists.
> nothing appears about its meaning.
>
> in the rdf 1.2 concepts document, it appears three times.
> two appearances are in diagrams.
> the third appearance is limited to a text passage which describes just
> that the term appears as the predicate in a reifying triple, but does not
> describe what that means.
>
> the rdf 1.2 primer includes the term once as the special predicate of a
> reifying triple.
>
> nothing in any of those documents describes what the a reifying triple
> means.
> that is, nothing defines how these two new features - propositions and
> rdf:reifies, are to be used to provide the capability set out in the
> charter: to provide the feature formulated by the rdf-star community group,
> to make statements about asserted statements.
> there has been correspondence to suggest that the rdf:reifies term is not
> even significant and that any predicate could serve an analogous role.
> without a specific meaning, there is no reason to include the term in the
> either the concepts or the semantics document.
>
> based on the current content of these documents i would not concur with a
> decision to advance them to candidate recommendation status.
>
> as alternatives, the working group could work towards some combination of
> - introduce a proper formal semantics for reifying triples as per pfps’
> sketch.
> - include a note in the semantics document to explain why the group
> refrains from that doing that.
> - remove the rdf:reifies term and related text from all documents other
> than the primer and formulate illustrative text in that document only.
> - approve the related prs to make the limits of the formalization clear
> and add a notes to explain the concerns.
> - include rdf:reifies, but without privilege rdf:asserts, along with, for
> example. rdf:mentions (i.e. weaker and stronger meanings, as pfps
> suggested)
>
> my concern is that these documents should serve to promote a shared
> understanding of how to work with the features which they introduce.
> throughout rdf history, that shared understanding has depended on formal
> definitions for the interpretations.
> while it may be true that the current documents provide a basis upon which
> one could formulate one, they provide none.
> they should not be promoted to candidate status as if they do.
>
> ---
> james anderson | james@dydra.com | https://dydra.com
>
>
>
>