Re: Comments to the latest RDF specs

At 12:18 PM 8/20/98 +0200, Juergen Modre wrote:
>Hello,
>
>Here a few little hints to the latest RDF specs.
>Sorry if I misunderstood anything.

Thank you very much for taking the time to send us comments.  And
please do not apologize for misunderstandings; at this stage of
the work, it is useful for us to identify any areas that may cause
such misunderstandings so that we can try to improve the document.

>RDF Model and Syntax Spec (20 July 1998)

very good -- and a reminder to everyone: please identify which
version of the document you have reviewed.  In this case, a
new draft was issued just before this mail was received here --
so a greater possibility for confusion exists.

>In Chapter 3.3:
>In other words, what is the object the statement is referring to.
>-> "?" at the end is missing :-)

oops; thanks.  Actually, my English grammar teacher would tell me
that the sentence is improper in more fundamental ways.  We'll
fix this in the next draft.

>RDF Schema Spec (14 August 1998)
>===============

All your editorial comments are appreciated; we'll make sure the editors
have seen your comments.

>And a few questions:
>-------------------
>In the RDF Model and Syntax spec is said: "... Property names must always
> be associated with a schema".

yes, but there is possibility of confusion in this sentence.  The
intent was not to say that an RDF Schema *must exist* in which a
property name can be found, but rather to say that there must be
a binding between property names and an identifier that
unambiguously associates that name with some definition.  The
form of this definition is not mandated; i.e. it could be a text
file, an audio recording, an HTML file, or -- and most preferably --
a chunk of RDF/XML that uses the RDF Schema vocabulary to describe
the property type.

>How it the relation in the future between the RDF Schema and an actual
> RDF instance (is this the correct naming?) of
>that schema defined in terms of:
>Do I always need a RDF schema for a RDF instance (like a DTD for SGML) or
>is it not absolutely necessary "to have really one created" at the
>"required" URI to the schema (like a DTD is not absolutely necessary
> for XML)?

The association between an RDF instance and the schema or schemata
used by that instance is given through the XML namespace declaration.
The schema identifier is the "namespace name" as that term is defined
in http://www.w3.org/TR/WD-xml-names.

We will not mandate that an RDF schema exist at the URI which is the
schema identifier.  But we won't give out many 'Good Web Citizen'
awards to folk who don't create machine-interpretable (i.e. RDF)
schemas either :-)

>Will there be RDF processors in the future able to validate an RDF
> instance against the RDF schema like DTD's in XML?

we certainly hope so!

>If, what is the behavior in an error case?

It will depend greatly on the context, I expect.  A processor that
reads RDF instances for purposes of security-related metadata will,
I hope, discard anything that doesn't validate against its schemata.

A search engine (ala AltaVista or Lycos) may chose to ignore
invalid metadata or may index the metadata anyway but mark it
lower in ranking for retrieval purposes.  If the RDF instance
purports to use property types that are not defined by the
corresponding schema, it probably makes sense to exclude those
properties from the index.  But if the property exists and the
value fails some other constraint given in the schema then
maybe the index should still hold the property with an annotation
that its value is "suspect".

> Are there any good examples anywhere showing an RDF schema and examples
> corresponding to this schema in the RDF Syntax.
> Perhaps also with graphics illustrating it better.

Coming soon, we hope.  Many groups are working on interesting
RDF Schemas.  We will document them on http://www.w3.org/RDF/
as soon as we receive authoritative information that we can
release.

> BTW: Maybe such an example would be of interest to be included
> in the next spec.

Was the Dublin Core example (split between the RDF Schema and
RDF Model and Syntax specs) of any help to you?  If not, what
sort of example might you find more helpful?

-Ralph

Received on Thursday, 20 August 1998 16:35:32 UTC