Re: semantics and RDF(S)

> The problem that I see has to do with the meaning of constructs in RDF and
> RDF Schema.  Although RDF and RDF Schema are supposed to provide machine
> understandability for the WWW they have no semantics for most of their
> constructs.

> In the effort of producing DAML+OIL some of us have tried to provide a
> meaning for part of RDF and RDF Schema.  I have gone a minimalist route,
> and as part of the denotational semantics for DAML+OIL I have created
> some semantic mappings for RDF and RDF Schema.  Richard Fikes and Deborah
> McGuinness have created an axiomatic semantics for part of RDF and RDF
> Schema.  Neither of these semantics is anywhere near complete---they just
> provide enough to get (part way towards) a semantics for DAML+OIL.

I don't think things are in nearly as bad a shape as Peter suggests. 
First of all, the axiomatic semantics for RDF and RDF-S that Deborah and
I produced was done with the intent of axiomatizing all of the
constructs and the semantics for those constructs described in the RDF
and RDF-S specs.  We can debate the extent to which we have succeeded
completely in that attempt, but I would suggest that the document we
have produced is a useful basis on which to build, to discuss the
intended meaning of RDF and RDF-S constructs, and to discuss the
omissions or other problems in the specs themselves.

Specifically, I would recommend that suggestions for modifying or
further specifying the semantics of RDF and RDF-S be made in the form of
proposed changes or additions to the axioms.

I agree with Peter's "minimalist route" approach mentioned above.  We
have done the same in the sense that we have axiomatized what we
understand to be in the RDF and RDF-S specs and have resisted adding
further semantic constraints or interpretations to those constructs. 
Indeed, as mentioned in our document, we have not assumed a set theory
or set interpretation of classes nor have we assumed that properties are
relations.  Those assumptions do not seem essential to describing the
intended meaning of RDF and RDF-S.

One of the reasons I prefer an axiomatic semantics (i.e., a translation
into first order logic) for these languages is that such a semantics
enables one to constrain the possible interpretations of the language
constructs but does not require imposition of any specific
interpretation.  Although I am, in general, a proponent of precisely
specified semantics for representation languages, the Web and the
current set of significant open issues regarding the intended meaning
and use of knowledge expressed by markup languages is a context in which
I think it is wise to constrain rather than designate the acceptable
interpretations of these languages.  Hendler's "'a' rather than 'the'"
maxim applies importantly here.

> Producing a good semantics for the constructs in RDF and RDF Schema will
> not be easy.

Well, again I am not as pessimistic as you seem to be.  We already have
a great deal of the semantics specified, and I see no barrier at this
point to incrementally adding to that specification.  There are no doubt
difficult problems remaining, but I think it is likely to be the case
that the current semantics specifications are or soon will be sufficient
to support almost all of the uses to which RDF and RDF-S will be put in
the next few years.  Importantly, the world is not going wait for us. 
Hence, it is important for us to have useful descriptions of intended
meaning available on an ongoing basis.

> Comments on this are very welcome.  Even more welcome would be efforts to
> come up with a semantics for RDF and RDF Schema.

Again, I modestly suggest that we already have one in the form of a
translation to first-order logic, and I recommend that suggestions for
modifying or further specifying the semantics of RDF and RDF-S be made
in the form of proposed changes or additions to the axioms.

Second, I suggest that we as a group be specific about the unanswered
semantic questions in the current specs for RDF and RDF-S.  For example,
you mentioned several such specifics in your message (e.g., What does
equality for bags mean?).  I think a central role that we who are
attempting to articulate the semantics of these languages can play is to
provide feedback to the language designers in the form of specific
questions that need to be addressed in the specs and precise
descriptions of specific problems that we see in the languages
(hopefully with well thought out suggestions for removing them).

Richard

Received on Friday, 12 January 2001 13:55:47 UTC