Review semantics document (2002-12-13#7) - part 2

Continuing my action 2002-12-13#7, as minuted:
   http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Dec/0231.html
this is a further partial review of:
   http://www.w3.org/2001/sw/RDFCore/TR/WD-rdf-mt-20030117/
from section 3.4 to the end of the document.

(My review comments up to section 3.3 are at:
   http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Dec/0324.html )

Many of my comments are marked [Editorial], which are merely issues I draw 
to the editor's attention.

Some are marked [For discussion], which may or may not be problems with the 
document, and request wider review.

Two are marked [Error], which I think need to be fixed.

Although I've read through the Lbase and proof appendices, I can't claim to 
have fully understood all the details.

...

[Editorial, nit]
Section 3.4, para 3:
Says that "... L2V is defined ...", but earlier the document says in 
section 1.3 says "it is impossible to give a sharp definition of LV".
Suggest "... L2V is required ..."

...

[Editorial]
Section 3.4, table following para 4:
Nothing is said here about the interpretation of forms like:
   "sss"@lang^^ddd

(per http://www.w3.org/2001/sw/RDFCore/TR/WD-rdf-testcases-20030117/#literal)

Later, it is noted that language tags are ignored for all but rdf:XMLLiteral.

...

[Editorial]
Section 4.2, para 2:
States the rdfs-closure is obtained by adding the rdf-closure rules plus 
those given.  The subsequent text then goes on to duplicate some but not 
all of the RDF-closure rules (rdf:nil rdf:type rdf:List and rule 
'rdf1').  This seems mildly inconsistent style.

...

[Error]
Section 4.3, rule rdfD4 has a conclusion:
   aaa rdfs:subClassOf eee .
that I think should read:
   ddd rdfs:subClassOf eee .

...

[Error]
Appendix A, Lbase translation rules, 1st entry:
Should "prefixed with /" be "prefixed with \"?

...

[Editorial]
Appendix A, rdfs axioms:

The textual layout is confusing;  for a while I thought there was an axiom:
   (rdfs:Class(?x) and rdfs:Class(?y) and ((forall (?u) (?x(?u) implies 
?y(?u))))

...

[For discussion]
Appendix A, rdfs axioms:

The axiom:
   (rdfs:Literal(?x) and Lbase:String(?y)) implies rdfs:Literal(pair(?x,?y))
seems to suggest:
   rdfs:literal( pair('chat','en'),'fr')
and other oddities -- should this axiom be:
   (Lbase:String(?x) and Lbase:String(?y)) implies rdfs:Literal(pair(?x,?y))
and
   (rdf:XMLLiteral(?x) and Lbase:String(?y)) implies rdfs:Literal(pair(?x,?y))
?

[or maybe:
   (Lbase:XMLthing(?x) and Lbase:String(?y)) implies rdfs:Literal(pair(?x,?y))
for the second part.  Er, I guess that would belong under rdfD axioms.]

...

[For discussion]
Appendix A, first rdfD axiom:
The conclusion ?x(?y)=L2V(?y,?x) is puzzling to me, because from earlier 
axioms the result of ?x(?y) is Boolean, true if ?y is a ?x.  But I 
understood L2V to be a mapping from lexical space to value space, of which 
the latter may be other than Boolean.  Should that conclusion be:
   ?x(L2V(?y,?x))
?

You later say that use of a datatype name as a function name has a 
different meaning, but that seems to be ambiguous with use of a URI as a 
functor indicating that it is an rdf:Class (from the 2nd rdfs axiom).

(I've not read Lbase recently, so maybe I'm missing something here.)

...

[For discussion/editorial]
Appendix A:
Does Lbase distinguish between URIs as functions, as in rdfs:Class(?x) and 
other auxilliary functions such as rdf-member(nnn)?  I assume so, because 
it doesn't make sense that rdf-member(nnn) means that rdf-member is a 
class.  I find this aspect likely to be confusing.

...

[Editorial]
Appendix B:
For consistency (of style), I think this should be marked "Informative".

...

[Editorial]
Appendix B, para preceding "skolemization lemma":
Having been involved with this group for a while, I think I understand what 
you mean by "skolemization of a query", and why it represents a different 
query.  But is this a common usage that would be understood by anyone who 
knows something of Skolemization?

...

[Editorial]
Appendix B, proof of "RDF closure lemma":
What is this HP that pops up?  I don't see a definition.

...

[Editorial]
Appendix B, proof of "RDFS closure lemma":

The terminating comment "(sketch) QED" seems a bit odd to me.

...

[Editorial]
Appendix C:
Is this informational or normative?  I'm not sure, though I don't recall 
the normative parts of the document *depending* on definitions given here.

...

[Editorial]
Appendix C, Class:
The phrasing here "as their rdf:type" seems to hint this is the resource's 
only rdf:type.
Suggest "as an rdf:type".

...

[Editorial]
Appendix C, Logic:
Is logic truly limited to a formal *language*?  I have always thought it 
also encompassed the ways in which the language can be used.
Suggest:  "A formal language which expresses propositions, with rules for 
expressing relationships between propositions so expressed;  also, the 
study of such languages."

...

[Editorial]
Appendix C, Ontological:
This is a term whose definitions never meant anything to me until I started 
to understand something of ontology from its practical uses.  For myself, 
understanding advanced when I got the idea that it was like applying data 
type like concepts to real world (non data) objects.

I was going to make a suggestion, but I'm sure it will be in some sense 
wrong; e.g. "Concerned with kinds or classes of things, and relationships 
between them".  I'm not sure what to suggest, but if the terms based on 
'ontology' are to be introduced, I can't help feeling that there is little 
benefit unless the introduction goes usefully beyond the bare dictionary 
definition;  e.g. the Concise Oxford Dictionary has "ontology: the branch 
of metaphysics dealing with the nature of being", which did nothing to 
enlighten me.

...

[Editorial]
Appendix C, Proposition:
Suggest: "A statement or expression that is true or false".

...

[Editorial]
Appendix C, Resource:
The definition given here seems to be very particular to its use with RDF, 
in a glossary which contains many terms that are drawn from much wider 
usage.  Suggest:  qualify the description in some way, e.g. "As used with 
RDF, ...".

...

The end.

#g


-------------------
Graham Klyne
<GK@NineByNine.org>

Received on Thursday, 2 January 2003 12:25:10 UTC