Announcement: RDF Schema Proposed Recommendation to undergo 2nd AC Review

[The following message was sent to W3C Members today.  I forward it
 to the public archive by permission of Tim Berners-Lee.  -Ralph Swick]

Dear Advisory Committee Representative,

After considerable deliberation and consultation with individuals in the
RDF community and the XML community, I have decided to submit the RDF
Schema Proposed Recommendation to a second Advisory Committee review to
consider some specific comments made during the first review. This
second review will begin after the completion of an architectural
briefing document, described below, that I am requesting from the
Metadata Coordination Group.

The vast majority of the review comments received on the RDF Schema
Proposed Recommendation (http://www.w3.org/TR/PR-rdf-schema) were in
favor of publishing this specification as a W3C Recommendation. Some
comments requested a few wording changes to clarify the relationship
between this work and the XML Schema work. The editors will incorporate
those suggestions into the specification before it is resubmitted. A few
comments expressed the wish that RDF Schema would include more
facilities for defining constraints on property values but agreed that
the specification as proposed was a useful basis on which to build.
Eleven Members responded that they intend to provide products that
implement or use RDF Schemas. One review acknowledges agreement with one
of RDF's overall goals of enabling the description of "assertions", but
disagrees with the solution adopted in RDF because it does not encompass
all of the XML infrastructure.

It is clear that the RDF Schema Working Group has met its charter and
produced a specification that is technically sound and sufficiently
complete. No technical issues were raised that require reconvening the
RDF Schema Working Group to address. The RDF Schema specification
provides the necessary core infrastructure for the metadata community to
begin defining interoperable machine-understandable vocabularies and to
allow implementation and deployment work to move forward. The RDF Model
and Syntax Recommendation (http://www.w3.org/TR/REC-rdf-syntax)
introduces the core data model and architectural philosophy of RDF; this
specification applies that model to the description of RDF vocabularies.

	XML-Schema and RDF

However, two reviews from W3C Members who participate in the XML Schema
Working Group suggest that the final XML Schema specification could
include all of the functionality present in the RDF Schema Proposed
Recommendation. Their suggestions warrant serious consideration: it
would be good to have a common architecture for tree-structured
documents and directed graphs.

If the suggestion is correct that the XML Schema specification can
incorporate the functions of RDF Schemas, and can be produced in a
timely manner, the Web will benefit from a consistent schema
architecture encompassing both syntactic and semantic information. Even
if the XML Schema document does not end up incorporating the RDF
functionality, it seems that a clear connection between the XML and RDF
architectures at the lower layers would be desirable. This will require
a certain amount of cooperation.

I am asking the Metadata Coordination Group, working with the XML
Coordination Group, to produce a briefing document outlining a possible
architectural relationship between RDF and XML data modeling. The
briefing document will accompany a specific set of questions that I will
ask you to consider during the second review. A reasonable target date
for this briefing document to be completed is the XML Activity
face-to-face meetings (http://www.w3.org/XML/Group/#sc23lk4j2) scheduled
for September 27.

There is a need to define a simple connection between the XML and RDF
architectures so that work may proceed in all communities that want to
use XML to distribute data on the Web without fear of later
incompatibility. I expect this work to have negligible impact on the RDF
data model. As the current RDF Schema specification is a simple
application of that data model, implementations written today to the RDF
data model should be easily adaptable to a different expression of the
schema data model. A closer architectural correspondence may have
greater impact on the RDF/XML syntax. One possible result may be a
simpler XML syntax for RDF than the present one, a result I am sure the
metadata community will also appreciate. Dan Connolly, Ralph Swick, and
I have written a note, Web Architecture: Describing and Exchanging Data
(http://www.w3.org/1999/04/WebData), describing our further views on
this relationship.

	Continuing RDF-based implementation

The metadata community has done a commendable job, with RDF Schema, of
distilling a broad wish list of items into a minimum core of properties
that enumerate metadata vocabularies. Some review comments on the RDF
Schema Proposed Recommendation expressed the wish that this minimum set
were a little larger, in particular that the RDF Schema language should
contain more constraint vocabulary for property values. However, when
reviewing their work in the context of new work items started within the
document markup activity, I reach the conclusion that this would be
inappropriate. The RDF Schema working group appropriately deferred
design choices that might tie the hands of the just-started XML Schema
work in the expectation that synergy with related XML activities may
allow the community to adopt common solutions to similar problems. The
metadata community has presented a proposal for the minimum vocabulary
set that permits unambiguous declaration of the existence of individual
metadata properties. This set is an essential piece of the Semantic Web
(http://www.w3.org/DesignIssues/Semantic). By taking advantage of the
RDF data model, the schema framework of the RDF Schema Proposed
Recommendation can accommodate other property constraints developed in
the future.

The metadata community needs, and the RDF schema document provides, a
vocabulary for describing the properties used to describe objects on the
Web. Other W3C activities (e.g. CC/PP and P3P,
http://www.w3.org/TR/1998/NOTE-CCPP-19981130 and
http://www.w3.org/TR/1999/WD-P3P-19990407/syntax respectively) and some
Submissions (e.g. UCLP and PIDL -
http://www.w3.org/TR/1999/NOTE-uclp-19990120 and
http://www.w3.org/TR/1999/NOTE-PIDL-19990209 respectively) can use this
vocabulary to produce machine-understandable descriptions of useful
parts of their specifications. Therefore, as the review of the RDF
schema specification uncovered no technical issues within the document
itself, we are at a stage where we encourage the community to use the
properties as proposed by the RDF Schema Proposed Recommendation and
report implementation experiences before the specification moves
forward. Feedback on implementation experiences with the RDF Schema
vocabulary should be sent to the RDF Schema WG mailing list
(mailto:w3c-rdf-schema-wg@w3.org - archived at
http://lists.w3.org/Archives/Member/w3c-rdf-schema-wg) or to the public
RDF comments mailing list (mailto:www-rdf-comments@w3.org - archived at
http://lists.w3.org/Archives/Public/www-rdf-comments).

Tim Berners-Lee, Director, W3C
Ralph R Swick, Metadata Activity

Received on Friday, 18 June 1999 15:30:14 UTC