Some general MT nit-picking

Some comments on the 28-Aug MT draft.  I think this is great, and highly 
edifying.  My comments mostly point up areas I was struggling to understand 
what was being said...

--

Secetion 0/general.


The Qname prefixes rdf and rdfs are used extensively.  For completeness, 
the corresponding URIs should probably be noted somewhere;  e.g. in section 0:

[[[
This memo uses the QName syntax to refer to certain URIs defined by 
RDF.  The following namespace prefix and URI values are assumed throughout:
   Prefix: rdf, namespace URI: http://...
   Prefix: rdfs, namespace URI: http://...
]]]

--

Section 1.


Although it becomes fairly clear on careful reading, it wasn't immediately 
obvious to me that this section did not deal with anon nodes.  Maybe a 1st 
sentence:

[[[
This section describes the interpretation of ground RDF graphs.
]]]


This section seemed to introduce symbols and notation without always 
clearly indicating what they were intended to denote:
   I
   vocab(I)
   <s p o>


5th para:  I take it the discussion of time-varying referents is not 
intended to preclude resources with time-varying attributes?  (E.g. a 
vehicle whose current position can be a function of time.)


Introduction of E and I(E):  "... RDF graph E in I ..." took several 
attempts to grok because I kept reading the "in" as set inclusion;  maybe 
"according to" or "under"?  Also, I found the description of E as an RDF 
graph to be slightly confusing, because some of the interpretations given 
were for things that I don't understand to be graphs:  <qLiteral>, <uriref>.


Denotation of I(E):  missing bracket I(s) in
   if E is an asserted triple with the form <s p o>
     then I(E) = true iff <I(s),I(o)> is in IEXT(I(p))...
(Isn't the 'iff' overkill here, in the presence of the otherwise clause?)


--

Section 2.

The meaning of B in "for some value of B defined on anon(E)" was not 
immediately clear to me.  I think you mean something like "for some mapping 
B from anon(E) to the domain of I".

Paragraph discussing treatment of anon nodes as existentially 
quantified.  "... there is no need to specify the scope of the quantifier 
within a graph."  Should that be "quantification"?  (I understand 
"quantifier" to be a variable name used to denote quantification.)

Next para:  I struggled with this;  "the same interpretation ... also 
assigns truth values to graphs with anonymous nodes", etc.  I don't see how 
this works.  The Interpretation defined for ground graphs seems to depend 
on having a direct model theoretic interpretation for the nodes mentioned 
(if E is <s p o>, I(E) = true iff <I(s),I(o)> is in IEXT(I(p))... )

--

Section 3.

When you say:
   I(rdfs:ConstraintResource) is in IC
do you mean ... is in IR?

The next condition is well-mangled by my browser;  is that the intersection 
of "ICEXT(I(rdfs:ConstraintResource)) and IP" ?

--

That's all for now.

#g



------------
Graham Klyne
(GK@ACM.ORG)

Received on Wednesday, 5 September 2001 10:25:46 UTC