Re: Schema comments

Graham,

Thanks for this! A lot of fair points, a few of which (the typose at
least) may have been addressed in the now-published WD already. I'm off to
Hawaii for WWW2002 shortly, and will be taking a little break from RDFS
the next 2-3 weeks, but I do think you raise a number of points worthy of
discussion. The notion of restriction in RDF needs better presentation.
This is connected to the removal of the 'constraint' vocabulary from RDFS.
I believe range and domain do restrict the interpretations of the RDF
graph (in the MT sense), but they can't be read as restricting in the XML
Schema  / DTD validity sense. That wasn't very articulate! I'll try to
come up with some words for this while on the road...

thanks again,

Dan

On Tue, 30 Apr 2002, Graham Klyne wrote:

> I've been looking through the schema draft, but am out of time before I
> could finish.  Here's what I have:
>
> Overall structure
> -----------------
>
> This is generally fine, but I found a couple of section headings confusing,
> and think there might be a couple more useful bits of introductory material.
>
> Currently, we have:
>
> Introduction
> Example
> RDF Schema
> Domain and Range Vocabulary
> RDF Schema Overview
> ...
>
> I'd like to see the 3rd para of "RDF Schema" in a section headed RDF
> Classes, and the section "Domain and Range Vocabulary" named something like
> "Application use of RTDF Schema".
>
>
> Introductory material
> ---------------------
>
> I also think there's some ideas that could usefully be spelled out in the
> introductory section "RDF Schema", along the lines of:
> [[
> The basic RDF model uses Resources and Properties to construct assertions
> [ref model theory].  RDF Schema introduces the idea of a RDF Class, and a
> vocabulary for describing certain aspects of RDF Properties and Classes.
>
> Because RDF Schema uses the RDF information model and RDF language,
> applications may define their own vocabularies to augment RDF schema
> descriptions of properties and classes.  (For example, CC/PP uses RDF
> properties to describe attributes of a client device.  It has been
> suggested that CC/PP might define some additional properties that apply to
> such properties to indicate how certain kinds of conflict are resaolved by
> CC/PP:  such properties might reasonably be viewed as an extension to the
> schema vocabulary.)
> ]]
>
>
> Schema as restriction vs schema as additional information
> ---------------------------------------------------------
>
> [This echoes some comments I've made with respect to the Primer...]
>
> I find the use of the term "restrict" in relation to RDF schema to be
> contrary to my understanding.  Adding RDF schema information doesn't
> specifically restrict the things one can say;  rather, it allows an
> application to deduce further things from what is said.  There is no
> meaningful concept of RDF schema-validity - adding RDF schema statements to
> a graph cannot somehow make that graph invalid (even though they may mean
> that some graphs are not meaningful in terms of the intended interpretation
> of the vocabulary used).
>
> E.g. Paragraph 3 in "Introduction" says "... to restrict the possible
> combinations of classes and relationships, ...".  I think that is wrong.  I
> don't think RDF schema imposes *any* such restrictions on correctly formed
> RDF expressions.
>
> Other similar occurrences:
> "Domain and Range vocabulary", 1st para, "...limitations on..."
>
>
> Section headings
> ----------------
>
> Adding section numbers would make it easier to make references into the
> document.
>
>
> Details
> -------
>
> "Introduction", para 4:
> "... types of relationships ..." is this an unfortunate use of
> "types"?  Suggest deleting "types of".
>
> "Introduction", para 4:
> "... allows for a finer grained mixing ...":  finer than what?   I'm not
> sure what this means, so can't really suggest an alternative.
>
> "Introduction", para 5:
> typo: "vocabualry"
> (Otherwise, I think this para is great, especially with respect to my
> comments above about "restriction".)
>
> "RDF Schema", para 1:
> I struggled to grasp the first sentence.  Maybe you meant something like
> "... , which in turn describe application-specific ..."?
>
> "RDF Schema", para 1:
> Reference to "core vocabulary":  I think this could be confused with RDF
> core, and would suggest referring to it as "schema vocabulary".
>
> "RDF Schema", para 2:
> Typo: "seekback"
>
> "RDF Schema Overview", table of RDF classes:
> "The class Resource":  I suggest "The class of all RDF resources".
> Do we need to be clear about whether this includes lexical values
> (denotations of literals)?
>
> "RDF Schema Overview", table of RDF classes:
> "The concept of Class":  I suggest "The class of RDF resources that
> represent RDF classes".
>
> "RDF Schema Overview", table of RDF classes:
> "The concept of a property":  I suggest "The class of RDF resources that
> represent RDF properties".
>
> ...
> This is as far as I got.
>
> I'm out of time for now - sorry!
>
> #g
>
>
> -------------------
> Graham Klyne
> <GK@NineByNine.org>
>

Received on Tuesday, 30 April 2002 19:17:13 UTC