Re: Reification of Sets (of RDF Statement, for Queries)

Brian McBride wrote:
> Sandro Hawke wrote:
> > Issue 1: RDF M&S Does Not Provide Sets
> 
> Hmmm, that looks more like an observation than an issue :)
> 
> Seriously though, I'm happy to capture this in the issues list.  It
> would be helpful to know why the absence of a set construct is a
> problem.  There are many things that can be added to RDF.  What is the
> argument that sets should be built in to the core?  Or is the problem
> that you feel that RDF as it stands cannot be used to describe sets?

Short Answer:
  My concern is that a decent set of collection abstractions
  should provide for sets.

Does that mean it's in "the core"?  It seems to be we have three
general areas of design effort and potential standarization:

    - the RDF model (aka the semantic web information model)
    - syntaxes (rdf/xml, n3, ...)
    - vocabularies (also known as: modules, libraries, namespaces,
                    ontologies, ... )

Within the area of vocabularies, I imagine W3C will standardize on
vocabularies for certain foundational subjects, including

      + numbers
      + characters
      + collections
      + RDF statements  (ie reification of RDF in RDF)
      + documentation  (rdfs:label, rdfs:description, ...)
      + a description logic  (daml)
      + a relational logic   (as SQL is based on)
      + a full first order logic

Within collections, it seems pretty clear to me that we ought to have
an abstraction for sets as used in set theory.  At very least, I think
the logics would like it.

That said, I'm not sure what "the core" is.  I think this is all a lot
like a programming language and its standard library or libraries.
The RDF Model is the programming language itself.  Unlike most, this
one has multiple equivalent syntaxes.  And the vocabularies are just
libraries, one or more of which is well enough made and useful enough
that people have little reason to re-invent it.  (Still, W3C will
probably make a 2.0 some day, so one should never think of it as The
Core Vocabulary.)

I worry that the new RDF working groups do not match this factoring of
the problem space very well; the charter of your group (RDF Core)
covers parts of all three areas, and that's likely to confuse people
about which issues are really part of the rdf/xml syntax, or some
vocabulary, or the basic model.

    -- sandro

Received on Friday, 13 April 2001 10:12:13 UTC