Editorial comments on RDF-based semantics

RDF-based semantics

I have only read a small part of this document.  I am concerned about
the clarity of the recommendation so my comments are
mostly editorial.

As I think another reviewer said, I'm struck by the inconsistent use
of "OWL 2", which appears not to refer to anything in particular. I
think this is confusing. The structural specification, direct
semantics, and RDF mapping, and XML serialization only apply to the DL
subset of the language (structural specification), and the document
titles and text should reflect this. If you really want to exclude
what's now called OWL 2 Full from qualifying as bona fide OWL 2, then
please be very clear about this, either by changing terminology or by
stating up front the exact relation between OWL 2 Full and OWL 2.

For example: this document says

   This document provides the RDF-compatible model-theoretic semantics
   for OWL 2, called "OWL 2 Full".

while the abstract of SSFSS says

   The ... structural specification ... provides a normative abstract
   model for all (normative and nonnormative) syntaxes of OWL 2.

It's hard to make sense of this.  The RDF-based semantics gives a
semantics of RDF graphs, only a subset of which correspond to OWL 2
(DL), so of OWL 2 = OWL 2 DL this would only make sense with the
qualifier "among other things" and with reference to the SS-to-RDF
mapping.

-----

It took me a while to determine that the function of the document is
to define what it means for something to be an OWL 2 Full model of
an RDF graph.  The title and introduction do not help readers to
understand this.  I suggest a title change, perhaps "OWL 2 Full
semantics of RDF graphs", and that the introduction should come out
and say something like this.

Going over the first few paragraphs:

   This document defines the RDF-compatible model-theoretic semantics of
   OWL 2, called "OWL 2 Full".

I think what you really mean here is semantics of "RDF graphs".  And
are you sure that you want "OWL 2 Full" to be the name of a semantics?
That seems OK to me, but it's sort of weird.  In common use I think it
will be taken to mean a language consisting of a combination of syntax
(RDF in any of its serializations) and
semantics (conditions on interpretation of the OWL 2 vocabulary).

Of course, as a side effect of giving the semantics of graphs, you
obtain an alternative semantics for OWL 2 (DL), via the reverse
mapping.  But that is a separate point.

			      The semantics given here is the OWL 2
   semantic extension of RDFS [RDF Semantics].

wouldn't it be better to say "OWL 2 Full semantic extension"?  It  
gives a
meaning to many graphs not in the image of the reverse mapping (IIUC).

					       Therefore, the semantic
   meaning

"semantic meaning" is redundant - there is no such thing as a
non-semantic meaning.  I think you mean one of "meaning", "semantics",
"formal meaning", or "formal semantics".

	  given to an RDF graph by OWL 2 Full includes the meaning given
   to the graph by RDFS. Beyond that, OWL 2 Full gives additional  
meaning
   to all the language features of OWL 2,

By "OWL 2" you mean OWL 2 DL, as defined in the structural
specification?  A hyperlink to a definition of "OWL 2" would help.

					 by following the design
   principles that have been applied to the semantics of RDF.

   OWL 2 Full accepts every well-formed RDF graph [RDF] as a
   syntactically valid OWL 2 Full ontology, and gives a precise semantic
   meaning to it.

Instead of "accepts" may I suggest "gives a meaning to". Otherwise one
is left wondering what kind of thing OWL 2 Full is that it's able to  
accept
or reject graphs.

I have no idea what an RDF graph that is not well-formed would be.
The cited document uses "well-formed" in several different ways, none
of which is what I think you mean.  Please delete all occurrences of
"well-formed" from this document unless you can provide or cite a
particular definition.

RDF graphs (which include all OWL 2 Full ontologies) are not
syntactic, and therefore cannot be syntactically valid.

Please use "ontology" consistently across all documents.  It has a
formal meaning defined in the SSFSS document which does not apply to
uses in this document.  Either change the term used in SSFSS, or
change or remove the term used in this document.

You don't exactly define "OWL 2 Full ontology" but I think what you
mean is an RDF graph that is intended to be interpreted according to
OWL 2 Full.  Since "intended" is not operational this seems like a
dubious term, and you should consider just saying "RDF graph" instead.

		 The semantic meaning is determined by the set of OWL 2
   Full semantic conditions, which include and extend all the semantic
   conditions for RDF and RDFS specified in [RDF Semantics].

What is the difference between "include" and "extend"?

See above re "semantic meaning"

							    OWL 2 Full
   acts as a vocabulary interpretation for the RDF and the RDFS
   vocabularies, and for the OWL 2 Full vocabulary. The OWL 2 Full
   vocabulary is a set of URIs that occur in the sets of RDF triples,
   which define the RDF syntax of OWL 2 [OWL 2 RDF Mapping].

Again I think talking about the "syntax" of an RDF graph is dubious
and unnecessary.
							    The OWL 2
   Full semantic conditions specify exactly which triple sets are
   assigned a specific meaning, and what this meaning is.

Why introduce the term "triple set"?  I think you're still talking
about RDF graphs here.

"the RDFS universe" -- this is not defined anywhere, not even in a
cited document.  The RDF semantics recommendation is very careful to
be "semantically neutral" and only talks about "a universe", not "the
universe".  The word "the" is distracting (which universe? what's in
it, exactly?) and not in accord with my understanding of how
model theory is supposed to work.

----

Section 6.3: The word 'valid' is used but I can't find a definition
for it anywhere.  By the SSFSS, "ontologies" by definition have a
certain amount of validity.  If used formally it should be
hyperlinked; if used informally I think it is unnecessary and should
be deleted.  "ontology" is defined formally and should be hyperlinked.

Also in theorem 6.1, I think you mean structural specification, not
functional-style syntax, since the reverse mapping produces the
former, not the latter.

Received on Saturday, 31 January 2009 16:10:50 UTC