Minutes 2001-11-16

RDFCore WG minutes for the Telecon 2001-11-16

(attached to this message)

1: Allocate scribe:  Graham Klyne

2: Roll call

    - Brian McBride (chair)
    - Daniel Brickley
    - Eric Miller
    - Dave Beckett
    - Jeremey Carroll
    - Jos De Roo
    - Jan Grant
    - Graham Klyne
    - Frank Manola
    - Aaron Swartz
    - Sergey Melnik
    - Dan Connolly
    - Pat Hayes
    - Ron Daniels
    - Mike Dean

    - Martyn Horner
    - Stephen Petschulat
    - KWON Hyung-Jin
    - Patrick Stickler

    - Frank Boumphrey
    - Bill dehOra
    - Rael Dornfest
    - Yoshiyuki Kitahara
    - Ora Lassila
    - Michael Kopchenov
    - Satoshi Nakamura
    - Pierre G Richard
    - Guha

3: Review Agenda


4: Next telecon - 10am Boston time, 30th Nov 2001

(As 22nd is Thanksgiving holiday in the US, will not have a telecon next week.)

5: Review Minutes of 2001-11-09



6: Volume of mail traffic
There has been a noticable increase in the volume of mail traffic to the
point where some WG members are feeling somewhat overwhelmed.  Is this
a widespread problem?  What can we do about it?

The biggest problem is that our chair is unable to keep up with the volume 
of traffic.  After some discussion, no single obvious solution, but 
participants are requested to consider any or all of the following:

- ensure subject line is meaningful.
- raise message priority for essential WG summaries, agenda, minutes, 
resolutions, actions, and other essential WG business items, to distinguish 
them from general discussion.
- Take conversations to private email, IRC or telephone (but post the 
summary of any outcomes back to the list).

7: Status of 2001-10-19#2 JanG  produce proposal on Entailment tests and 
test Manifest for 26/10/2001


8: Status of 2001-10-19#3  Jos Create test cases for model issues resolved 
at f2f


Status of 2001-08-02#9 2001-08-02#25  2001-08-02#26 2001-08-02#27
2001-08-02#28 2001-08-02#29 2001-08-02#30 2001-08-02#31 2001-08-02#32
2001-08-02#32 2001-08-02#17 2001-08-02#19 Danbri Fold agreed changes into 
RDF Schema WD

ACTION: 2001-11-16#1, DanBri, mail Brian with list of issues that have been 
folded into document.
ACTION: 2001-11-16#2, Brian, new test case from Jos to go onto futire 
telecon agenda.
ACTION: 2001-11-16#3, Jos, Send pointer to new test case to Brian.

10: Primer status

ACTION: 2001-11-16#4, Eric, Have updated copy available to group by 5PM EST 
Monday 19-Nov-2001.

11: Model Theory WD status

This has been on hold, pending resolution of dadatyping issues.  Resolution 
does not appear imminent.

ACTION: 2001-11-16#5, Pat, Post updated copy of model theory, with datatype 
discussion elided, for group review by Wednesday 21-Nov-2001.

12: Issue: rdf-equivalent-representations

   o The WG agrees that:

       - the graph model which is the basis for the model theory
       - the n-Triples representation of an RDF graph
       - the diagrams of graphs used in documents such as the RDF Model
         and Syntax document

      are currently all equivalent

   o The WG resolves to maintain that equivalence.

   o The WG notes that the RDF/XML syntax as currently defined is unable
     to represent an arbritary RDF graph.  In particular, the RDF/XML syntax
     cannot fully represent a bNode which is the object of more than one

   o The WG believes that extending the RDF/XML syntax so that it can 
     all RDF graphs is beyond the scope of its current charter and resolves
     to postpone consideration of this issue.

   o The WG actions the editor of the RDF Syntax WD to include in that
     document a clear statement of the RDF graph structures that RDF/XML is
     unable to represent.


RESOLVED with 1 abstention.
(With the caveat that equivalence is a statement of intent rather than 
certified fact).

13: Issue: rdfms-assertion

   The WG resolves that
     o the RDF model theory defines the semantics of RDF
     o it is beyond the scope of this WG to decide on how the laws
       of different countries should apply to statements made in RDF.
     o this issue be closed.


The group were not able to close this issue at this time, due to lack of 
agreement about what it meant to be sustainable in law, and the extent to 
which the WG could offer such assertions.

ACTION: 2001-11-16#6, Aaron, to resubmit alternative proposal based on MIME 
type for further discussion.

DanC provided a pointer to Aaron's proposal in IRC:

GK also requested that points raised in here be considered by any proposal:

14: Issue rdfms-boolean-valued-properties

   o The WG notes that rdf:type can be used in place of the rdf:is suggested
     in this issue.  Thus, to say that a resource is a chocolateLover,
     the class ChocolateLover can be defined, and the resource declared to
     be a member of that class using rdf:type.  rdf:isNot can be represented
     by declaring the resource to be a member of the class of
     NotAChocolateLover, again using rdf:type.

   o The WG resolves to close this issue on the grounds that current definition
     is adequate.


The use of rdf:type to express a Boolean predicate was not controversial, 
but the participants couldn't agree what to say about negation (rdf:type 
ChocolateLover vs rdf:type NotAChocolateLover).  In particular, the extent 
that we should reference pending work from the WebOnt group.  Pat was 
concerned that adopting the negation aspect could break the model theory.

ACTION: 2001-11-16#7, Pat, following email discussion of this topic, 
prepare new statement of this resolution to bring back for approval.

15: Issue #rdfms-rdf-names-use
Propose the WG

   o resolves that the use of rdf:Description except as the name of a
     description element is an error
   o resolves that the use of rdf:ID, rdf:about, rdf:resource, rdf:bagID,
     rdf:parseType except as reserved names as specified in the grammar
     is an error
   o resolves that the use of rdf:Bag, rdf:Alt and rdf:Seq except as typed
     nodes is an error
   o resolves that the use of rdf:li as a typed node is an error
   o resolves that the use of a container membership property (rdf:_nnn) as a
     typed node is an error
   o resolves that test case
     be obsoleted
   o resolves that  a copy of that test case be created as an error test case
   o actions DaveB to create test cases for the above cases
   o actions DaveB to identify any similar cases to those above and create
     test cases to cover them also


This item was not discussed.

16: Issue rdfms-aboutEach

   o  the WG resolves to remove rdf:aboutEach from the language on the grounds
     - it is not used
     - it is not widely implemented
     - it has confusing interactions with bagID as recorded in
     - it prohibits the development of  streaming RDF/XML parsers
     - it requires schema processing in the parser
     - this is the wrong layer in which to implemenent such functionality

   o Action DaveB remove from the grammar in the RDF/XML document

   o Action Brian update the issues list, especially Attention Developers

This item was not discussed.

17: Datatypes
Take a straw poll of the WG's views on the various datatype proposals
and discuss how to proceed


The schemes considered were:

U - datatype+literal embedded in URI form;  uses special URI form.
S - denotation of intermediate bNode is data type value;  literals are all 
P - use rdfs:range to indicate data type;  this means different instances 
of a literal string can denote different values;  requires change to the 
model theory.
P++ - like P, but also allows type information to be attached to literal 
instances (with literals as subjects)
X - was not included in this staw poll because no advocate was present to 
explain it.

In straw poll, S was clearly the most favoured system (by 9), and nobody 
said they could not live with it.  P/P++ was second most favoured (by 3), 1 
could not live with it.

2 participants said they needed more time to review and understand the issues.

GK expressed concern about compatibility of S with existing RDF usage, esp. 
CC/PP;  it seems that existing instances are OK, but existing schema may be 
broken.  Or vice versa.  Can we do better;  e.g. per 
http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Nov/0534.html ?

Pat noted that proposal S is clearly not compatible with DAML+OIL's 
treatment of literals.  That proposal uses scheme P.

Frank noted that the broader implications of any proposal selected need 
further consideration.

For the time being, we will pursue proposal S, subject to review.

ACTION: 2001-11-16#8, Sergey, Get datatype discussion document to WG as 
soon as possible.

ACTION: 2001-11-16#9, Frank, Clarify the architectural and other broader 
concerns with any datatyping scheme that must be considered.

ACTION: 2001-11-16#10, Jeremy, raise compatibility issues concerning 
schemes S and P with the WebOnt working group.

ACTION: 2001-11-16#11, MikeD, raise compatibility issues concerning schemes 
S and P with the DAML+OIL joint committee.

18: Semantics of Reification - what progress

This item was not discussed.

***Meeting closed***

      /\ \    Graham Klyne
     /  \ \   (GK@ACM.ORG)
    / /\ \ \
   / / /\ \ \
  / / /__\_\ \
/ / /________\

Received on Friday, 16 November 2001 13:27:46 UTC