Re: Cambridge Communique

At 10:29 AM 11/10/1999 -0800, Perry A. Caro wrote:
http://lists.w3.org/Archives/Public/www-rdf-interest/1999Nov/0019

> How do we decide which one to use (RDFSchema vs. XSchema)?

This will certainly depend very much on what your application
requirements are.  In fact, I suspect a large number of applications
to be interested in using *both* languages.  The understanding that
we reached at the Cambridge meeting was that these two languages are
complementary, not mutually exclusive.  The XML Schema languages
permits expressions of some features of XML documents, the RDF
Schema language permits expressions of other features of data models.
XML Schema cares more about the form of the expression, RDF Schema
(and RDF Model&Syntax) care more about the meaning of the expression.

Many of us (in both RDF and XML "camps") hope that it will be
possible to use both languages together in a single document.
So the current confusion about multiple uses of the term "schema"
may then be seen to be a clever bit of foresight.

> Are we going to have to live in a world where we have to support both, and
> have two sets of processors, two sets of specifications, two sets of user
> guides, converters to/from, etc., etc.?

The kinds of processing one wants to do on *data*, an in particular
graph-structured data, will require some additional APIs beyond what
is wanted for tree-structured *documents*.  Exactly what those
general-purpose APIs are is part of what I hope this Interest Group
will eventually propose.  If a single D{ocument,ata} Object Model
processor can reasonably be understood to support both then we
ought to make that happen.  But multiple object classes that provide
different views on a single data structure is also a bit of technology
that we all understand how to work with in our toolset.

> my conjecture that the reason RDFSchema
> is so underspecified is because of the long shadow cast by the impending
> XSchema?

confirmed.  In fact, we made the very explicit decision to defer
most data typing mechanisms within the RDF Schema specification
in the hope that RDF Schema writers would be able to use the
XML Schema data typing language.  The RDF Schema Working Group
decided that RDF Schema was useful to a sufficiently large
community right away without detailed data typing and that
it would be a problem for a much longer period of time if these
two specifications wound up with similar but slightly different
basic data typing facilities.

> RDFSchema lacks some
> obviously useful features, like concrete syntax for specifying whether a
> value is required or optional, or read-only in a named context, etc. 

One of the primary goals of RDF was to provide a framework in which
communities could create vocabularies without having to wait for some
central agency, and to permit independently-developed vocabularies to
be freely mixed in a document with minimal confusion.  When later we
discover that there are mutiple terms for the same concept we add
that relationship to our RDF graphs too.

Semantics such as "required", "optional", "repeatable", etc. were
known to be desired within XML Schema too (since they exist in
DTDs) and, as above, the RDF Schema Working Group proposed to
defer adding this vocabulary in the interest of promoting easier
reuse of XML technology.  It is very encouraging to hear on this
list that others have indeed found the framework and tiny vocabulary
set of RDF Schema 1.0 to be useful, understanding that it can easily
grow as we identify common needs not handled elsewhere.

-Ralph Swick

Received on Friday, 12 November 1999 10:35:00 UTC