Meeting
Minutes
: 2002-08-10 : Montréal
Meeting
Secretary, Review and Publication: Bernard Vatant
Release: 2002, August 14.
Attending Members
Bernard Vatant, Lars Marius Garshol, Mary Nishikawa, Peter Flynn,
Holger Rath, Michael Priestley
Excused Members
Thomas Bandholtz, Suellen Stringer-Hye, Jim
Mason, Eamonn Neylon, Steve Pepper
New Prospective
Members
Vivian Bliss (Microsoft) and Don Smith (Isogen) attend the meeting
and request to become participating members.
Invited Guests
Patrick Durusau (SBL), Nikita Ogievetsky (Cogitech), Eric Freese
(Lexis Nexis)
1. Approval of
previous meetings minutes:
Approved
with no comment
2. Short Review
of current TC work and status:
This review seems necessary for everybody, both new and "ancient"
members ...
- Review of Charter
It is needed to tune the Charter with the actual current TC work and
orientation. A New Charter proposal has been already posted on the TC
Web Site, but more time is asked by members to review it.
It will be discussed and voted on-line.
- Structure of
Deliverables
The structure of deliverables proposed in Barcelona has been approved.
It is agreed that Deliverable 1 should and could be delivered ASAP (before
end of September)
3. Discussion of
Deliverable 1
- Introduction
Needs to be polished and be added some illustrating diagrams. Needs
also to be added some clear exposition of what will be addressed by
this deliverable and what issues will not (e.g. structure of identifiers
and indicators)
- Requirements
Consensus is that URNs should not be recommended as PS Identifiers,
default of a specific resolution process. Use of non-resolvable URNs
is ruled out by a new expression of Requirement 2, getting rid of the
condition "if the URI is an URL". The Requirements are now
expressed as following:
- Recommendations
The content and expression of Recommendations is revisited throughout,
focusing on Publisher and Metadata.
- Publisher:
long debate on what it means, if it is a specific information or if
it should be included in the general subject metadata. Consensus is
that the notion is critical and should be declared independently of
any other metadata, but that it is still ill-defined in the Web universe.
Decision
: stick to Dublin Core definition, which is not very precise but is
available and widely used standard.
A new definition in a similar context would lead to nothing but confusion.
- Metadata:
Make distinct recommendations for human-readable metadata, machine-processable
metadata, and their necessary consistency.
The whole recommendation
list is redefined as following (subject to re-wording and refining)
- Published Subject
Indicator should provide human-readable metadata
(included in resource or referenced from it).
- Published Subject
Indicator should provide machine-processable metadata
(included in resource or referenced from it).
- Human-readable
and machine-processable metadata should be consistent (but not necessary
equivalent)
- Published Subject
Indicator should indicate that it is intented to be a PSI.
- Published Subject
Indicator should identify its publisher.
- "Apples
and Oranges" example has to be purged of some recommendations
that do not belong to the scope of this deliverable, like those concerning
URIs "human-readable semantics" and use of the "PSI"
token.
- Specific syntax
given in the example should be well thought out.
- XTM is ruled
out at that stage, could be introduced by Deliverable 2. XHTML and
RDF are still in.
- The example should
also be explained in
terms of, or at least illustrated by, diagrams.
- The use of fragment
identifiers is still open. It is thought that they will be used anyway
(e.g. by GeoLang)
- It is agreed
that the editor (Bernard V.) should work out the example, since the
original author (Steve P.) seems not likely to have the bandwidth
to achieve this task.
4. Roadmap and
next meeting(s)
Bernard will propose
ASAP a new and complete draft for Deliverable on the basis of the above
decisions. This draft should be close to be the final one for this document,
with the aim of review and approval before end of September.
Next F2F at the end
of Baltimore XML 2002, on December 14th.
Conference Call(s) in-between to be decided on the basis of on-line work,
with a 3 weeks notice delay at least.
|