W3C home > Mailing lists > Public > public-rif-wg@w3.org > March 2007

RE: Glossary entry: Abstract syntax

From: Gerd Wagner <wagnerg@tu-cottbus.de>
Date: Sun, 11 Mar 2007 23:43:56 +0100
To: "'Dave Reynolds'" <der@hplb.hpl.hp.com>
Cc: "'Chris Welty'" <cawelty@gmail.com>, "'RIF'" <public-rif-wg@w3.org>
Message-ID: <001301c7642e$c1695f10$14b2a8c0@informatik.tucottbus.de>
>>> In a rule metamodel a function need not have a sort.
>>> It will have a sort, though, in the underlying vocabulary
>>> metamodel (which you still miss to discuss, btw).
>> I have no idea what you're talking about.
> Gerd - I could guess that by "vocabulary" here you might mean
> "library of builtin functions with their signatures" or you
> might mean "symbol table of user constants defined within the
> current scope along with their signatures".

Basically, yes, a vocabulary is a more explicit form of what is also called
a "signature" in predicate logic:
it defines the symbols for individuals, functions and predicates
(pre-defined/built-in and user-defined).

Also an RDF schema defines a vocabulary, a set of individuals and predicates
(classes and properties).
It's a credo of the business rule community that every ruleset has an
underlying vocabulary that has to be
provided along with the rule set. Only logic programming people tend to
ignore the role of the vocabulary, 
since it is left implicit in Prolog.

Concerning the RIF condition language, we should make explicit that the
elements of "Const" may be
data literals, object names, function names or predicate names. Although the
current document mentions
the possibility of an integer numeral denoting a predicate, I think we
should only allow a symbol overlap
among names for objects, functions and predicates as in the following
diagram (in fact, the only use 
case for a symbol overlap seems to be the case of properties which are both
functions and predicates).

Now, a RIF vocabulary metamodel will define at least three (possibly empty
and possibly overlapping) sets 
of URIs as symbol sets for objects, functions and predicates. It may look as
shown below. These 
declarations (corresponding to individual axioms, class axioms and property
axioms in OWL) have not
yet been included in the syntax definitions of the current core working
draft, although they will certainly
be needed. When looking at this vocabulary issue, you may also notice that
RIF did still not clarify in
which form (object) classes will be included: will they be special unary
predicates or will they be
special sorts?



(image/gif attachment: LogicalNames.gif)

(image/gif attachment: Vocabulary.gif)

Received on Sunday, 11 March 2007 22:44:07 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:33:38 GMT