Re: Statements/Reified statements

Pierre-Antoine CHAMPIN wrote:

> Well; I feel like starting a new thread about unicity of triples,
> my mailer windows is no longer wide enough :)
>
> The more I read the messages on the list and the specs,
> the more I think there is a possible unification of all our views;
> I'll try to sum it up, being consistent with the spec.
>
> First, Statements and Reified statements are not the same thing.

Agree.

> Statements are uniquely defined by their subject/predicate/object. The spec says so (section 5, item 4). They are therefore what we sometimes called "abstract statement". The spec does *not* say that they are resources.

Agree (with reservations).  That's the way mathematicians talk about things and that's how they become abstractions.  The Semantic Web, however, will not be an abstraction, it will be a collection of real statements each of which gains its unique identity (and its
meaning) from its context of utterance.

But there may be a way to fix [RDF data model] so that it is clear to which set of things we refer - the Platonic abstraction in the sky or the real things in our computers.  Simply change "The" to "A" in [S&M].   "The RDF data model" does refer to your mathematical
abstract set; but "A RDF data model" would refer collectively to our distributed semantic islands.   Then I can say "A triple is unique in a RDF data model" and the picky pedantic agents in MyMemory won't complain.

[RDF data model] http://www.w3.org/TR/REC-rdf-syntax/#model

> Reified statements are resources, defined in section 5, item 9. They *represent* a statement -- it sounds therefore fair that a statement may have more than one representation. They also have the 4 well know properties, now often called the "reification quad".

Agree.  I have asserted "reification quad" to MyMemory.

> We also defined a "stating" as the *expression* of a statement in some piece of RDF (i.e. an arc in the graph, a triple occurence in a model). It is clear that there can be more than one stating for the same statement (although there may be only one per graph/model).

Agree.

>   Pierre-Antoine  -- suggesting (c), but waiting other's opinions ;-)

I vote for (c) too.   We could even define property arcs that would disambiguify the interpretations in our models.  Example:  <rai:english id="english1">"I said Bush won the Election [stid1] , but may have jumped the gun; however after the Electoral College meets in
December, I will be able to say with certainty that the proposition   'Bush won the Election' [stid2] is true [stid6]."</rai:english>

stid1 [Bush] rai:wonThe: [Election]
stid2 [Bush] rai:wonThe: [Election]

[s1] rdf:type [statement]
[s1] rdf:subject [Bush]
[s1] rdf:property [wonThe]
[s1] rdf:object [Election]
[s1] rai:reifies: [stid1]
[s1] rai:reifies: [stid2]
[s1] rai:parsedFrom: [english1]

stid3 [s2] rdf:type [proposition]
stid4 [s2] rai:propositionOf: [s1]
stid5 [s2] rai:after: "December 17, 2000"
stid6 [s2] rai:is: [True]

Note that I am calling the abstract statement a "proposition".  I believe that is the correct term for it in formal logic.  Also note that [s1] is superfluous if we allow that stids can be objects of triples [StatementsCouldBeObjects].

[StatementsCouldBeObjects] http://lists.w3.org/Archives/Public/www-rdf-interest/2000Nov/0317.html

Seth Russell

Received on Wednesday, 22 November 2000 13:41:05 UTC