XML [SCL] Issues and proposals (details)

[Chris Menzel]

On Thu, Jan 16, 2003 at 12:18:31AM +0200, Tanel Tammet wrote:
> 
> Christopher Menzel wrote:
> >>This does not mean avoiding
> >>RDF, quite the opposite: IMHO a "semantic translation"
> >>should be into RDF, not SCL-XML.
> >
> >
> >But how is that possible given RDF's expressive limitations, Tanel?  
> >Perhaps I am not understanding what you mean by a "semantic translation".
> 
> A translation where we encode predicates RDF-style (also encoding
> N-ary predicates via a set of new predicates) and add our semantics to
> parts of the language (saying that some things are implications,
> negations, variables etc, 

OK, Tanel, I think I understand the idea.

> although they have no such "default" meaning in RDF itself).

OK, I think I understand the idea.

> although they have no such "default" meaning in RDF itself).

Much more than that, I think.  To say that they have no default meaning
in RDF suggests that the resources are present *in RDF* for making those
meanings explicit.  For example, there is no default meaning for, say,
"natural number" in FOL per se, but we can in an certain sense provide
one by adding a predicate symbol "N", a constant "0", function symbols
"s", "+" and "x" and, say, the axioms of Peano Arithmetic.  By contrast,
because of its expressive limitations, there is no way analogous way
simply to add, say, some new tags to RDF that would enable us in any
robust sense to capture the meaning of arbitrary propositions of a
first-order language.  RDF simply doesn't have the semantic resources;
we simply cannot "add the semantics" of full FOL to it.

This leaves me wondering (quite possibly purely out of ignorance) what
the purpose of such a "semantic translation" would be.  As I understand
the proposal, RDF would only be serving as a sort of metalanguage for
talking about a given first-order language L.  But it would be a curious
one, as it would mostly be serving simply as a sort of quoting mechanism
("Here is a quantified sentence of L"; "Here is a conjunction of L";
etc); it would be incapable of doing any of the sorts of things that a
more typical metalanguage can do (e.g., describe the syntax of a
language, define its semantics, etc).  The actual meanings of the quoted
sentences could only be captured by mapping them once again from the RDF
metalanguage into some framework having at least the expressive power of
L -- leaving me wondering what essential role the RDF middleman is
playing.

It is quite possible I am not understanding the purpose you envision
for this functionality, and that my concerns about expressive power are
simply moot.  I'd appreciate enlightenment on any such point!

Regards,

-chris

--

 /\ ASCII ribbon | Chris Menzel -- http://philebus.tamu.edu/~cmenzel
 \/   campaign   | Philosophy Dept, Texas A&M University
 /\   against    | College Station, TX 77843-4237
/  \ HTML email  | voice: 979.845-5660  fax: 979.845.0458