"Quick" review of vocabulary document

Reading through this, I noted a number of minor editorial nits, nothing too 
serious.  My comments here focus on what I perceive to be interactions with 
the formal semantics.

Reviewing document:
   http://www.w3.org/2001/sw/RDFCore/TR/WD-rdf-schema-20030117/

I don't think any of the points I raise are definite showstoppers, but I 
think those labelled "[for discussion]" should be reviewed by the editor 
before going to last call.

...

[Minor editorial]

Section 1, para 9:
[[
This specification does not attempt to enumerate all the possible forms of 
vocabulary description that are useful for representing the meaning of RDF 
classes and properties. [...]
]]

I think "representing" here should be "describing"  (to me, "representing" 
gets uncomfortably close to the formal concept of denotating, and I don't 
think that's what is mean here.

I suggest:
[[
This specification does not attempt to enumerate all the possible forms of 
vocabulary that are useful for describing the meaning of RDF classes and 
properties. [...]
]]

...

[For discussion]

Section 1, para 10:
[[
The language defined in this specification consists of a collection of RDF 
resources that can be used to describe properties of other RDF resources 
(including properties) in application-specific RDF vocabularies. [...]
]]

I don't think it's correct that the *language* consists of a collection of 
RDF resources.  I suggest:
[[
The language defined in this specification consists of a collection of URIs 
that denote RDF resources that can be used to describe properties of other 
RDF resources (including properties) in application-specific RDF 
vocabularies. [...]
]]

...

[For discussion]
+
[Editorial]

Section 2.3 (rdfs:Literal):

I don't think this section sits comfortably with the formal description of 
rdfs:Literal, which is much less definite about the instances of this 
class.  "The set of all possible values of all literals is assumed to be a 
set called LV. Since the set of datatypes is not restricted by RDF syntax, 
it is impossible to give a sharp definition of LV, but it is required to 
contain all literal strings, all pairs consisting of a literal string and a 
language tag, and all values of typed literals." -- 
http://www.w3.org/2001/sw/RDFCore/TR/WD-rdf-mt-20030117/

I'd suggest wording that is a little more obviously open-ended (see below)

+

[[
The class rdfs:Literal is the class of literal values such as strings and 
integers. Property values such as textual strings are examples of RDF 
literals. Literals may be plain or typed.
]]
Is the term "literal values" explained anywhere?  If so, then I suggest a 
cross-reference;  if not, I'd suggest a more explanatory wording:
[[
The class rdfs:Literal is the class of literal values (i.e. values that can 
be denoted by literals), including the likes of strings and integers.
]]

(Also note:  in my comments to the semantics document, I question whether 
there really is a distinction between the set of literal values and 
resources in general.)

(Also, as noted below, I am uncomfortable with the phrase "property value".)

...

[For discussion/editorial?]

Section 2.3, 3.1, and elsewhere I think:

There are a number of references to a "property value" or "the value of a 
... property" meaning the thing denoted by the object of a statement 
containing a given property.  When I see the term "property value", or 
similar, I tend to think of the denotation of the property itself.  It 
would be good to find a snappy phrase that didn't have this potential 
misinterpretation.

Maybe:  "property object value" or "the object value of a ... property"?

...

[For discussion]

Sections 3.6, 3.7:

It's not clear what are the implication for the domain of rdfs:label and 
rdfs:comment being rdfs:Literal.  I think some people might read this to 
mean that any occurrence of these properties must be syntactically with a 
literal object.  I don't think that is correct.

I think it would help to be clearer about this;  e.g.

[[
The range of this property being rdfs:Literal means that its object value 
must be something that can be denoted by a literal;  in normal use, the 
object will be a literal, but this is not a requirement and it is quite 
valid to use a non-literal resource that happens to denote some literal value.
]]

In all, I think the rdfs:Literal range constraint is semantically very 
weak.  My concern is that it is mistaken as a syntactic constraint.

...

[Editorial]

Section 5.3:

[[
The original RDF Model and Syntax Specification [RDFMS] defined vocabulary 
for describing RDF statements without stating them.
]]

I think the term "stating" here may be not understood.  I suggest:

[[
The original RDF Model and Syntax Specification [RDFMS] defined vocabulary 
for describing RDF statements without asserting them (in the sense of 
[x-ref to Semantics Glossary, "assertion"]).
]]

...

[For discussion]

Section 6.1:

rdfs:Literal -- see comments above.  I suggest the comment might be:
[[
The class of literal values, including values such as textual strings and 
integers.
]]

(I'm trying to offer wording that is more suggestive of an open-ended set 
of values.)

rdf:XMLLiteral -- I think the comment is not quite right -- it seems to 
admit any syntactically valid piece of XML;  I suggest:
[[
The class of XML literal values;  i.e. canonical XML (cf. denotation of 
"xxx"^^rdf:XMLLiteral in semantics document, section 3.1).
]]

...

Appendix A, not checked.



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

Received on Monday, 13 January 2003 05:12:25 UTC