- From: Ralph R. Swick <swick@w3.org>
- Date: Tue, 07 Sep 2004 11:27:20 -0400
- To: public-rdf-in-xhtml-tf@w3.org
RDF-in-XHTML Task Force
7 Sep 2004
[2]Agenda
See also: [3]IRC log
Attendees
Present
Ben Adida, Ralph Swick, Mark Birbeck, David Wood,
Masayasu [Mimasa] Ishikawa, Steven Pemberton, David Wood,
Beth Epperson, Mark Seaborne, Rich Schwerdtfeger,
Kevin Kelly, Shinichi Matsui, Jeremy Carroll,
Dan Connolly, Dom Hazael-Massieux
Chair
Ben Adida
Scribe
Ralph Swick
Contents
* Topics
Presentation of XHTML 2.0 Metainformation Module
Discussion
* Summary of Action Items
_________________________________________________________________
Previous: 2004-08-04
[6]http://lists.w3.org/Archives/Public/public-swbp-wg/2004Aug/0020.html
agenda
[7]http://lists.w3.org/Archives/Public/www-archive/2004Sep/att-0010/telecon-2.html
Steven: HTML WG plan is to go to Last Call the end of this month
MarkB: presents Metainformation Module
... recall motivations:
... trying to address 2 different worlds; HTML and RDF
... HTML wants to do what they do every day; e.g. metadata for news
... RDF wants to join metadata from different sites
... want to make it so people in HTML world are encouraged to put more
of the rich metadata into their documents
... and so people in RDF world can get this data out
... not a solution for HTML community to put metadata into an external
document
... but also HTML community want more metadata elements; couldn't
address these individually, wanted an open-ended solution
<Steven> (wanted more semantic-based elements -- would be closer)
Mark: need to keep in mind that we have these two communities
... original attempt was to try to import all the RDF attributes into
XHTML
... thought we had succeeded, such that a current RDF parser would be
able to accept an XHTML document
... but ultimately this did not succeed
... in the process we did have triples, properties, etc.
... current syntax is similar to ntriples
... does have nesting; e.g. the subject of a nested <link> can be
[inherited from] the nesting <link>
... then we evolved the syntax so the new attributes for <meta> and
<link> could be used anywhere else in the document
... this encourages authors to put more metadata in the document
... the attributes are the key part of the syntax
... we feel all of the base RDF concepts spec is covered by this
syntax
... may need a default datatype for <meta>
... e.g. all content of <meta> is an XML Literal
... not reasonable for a document author to have to say this in every
document
... RDF containers and collections are probably covered
... rev attribute should allow a list of items to be identified as a
Bag
... so I think we have pretty much everything in RDF [Core]
... responded to some email comments;
... e.g. normative reference to RDF
... HTML WG agrees; are looking for suitable wording
... the motivation for the syntax _is_ to provide RDF
Dom: I sent some quick comments several weeks ago
[[
Quick glance through xhtml2 meatainformation module
From: Dominique Hazaël-Massieux <dom@w3.org>
Date: Fri, 06 Aug 2004 11:33:20 +0200
To: public-rdf-in-xhtml-tf@w3.org
Message-Id: <1091784800.1416.2454.camel@stratustier>
]]
--
[8]http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2004Aug/0026.html
Dom: identified some potential confusion in the model between XML
fragments and strings
... lack of reference to RDF was damaging; not constraining the source
of [properties] will make it difficult to interpret of the results
Mark: HTML WG would like some feedback on normative RDF references
... personally am not sure what a normative reference would be
... was thinking specifically of the RDF Concepts specification
... in several places, "the RDF Spec" means the whole set of 6
documents
... so what does it mean to say "we are referring normatively"
Jeremy: a reference to Concepts should suffice
... saying you are generating an RDF graph as referred to by Concepts
... these graphs are intended to be used according to RDF Concepts
<dom> -> [9]http://www.w3.org/TR/2004/REC-rdf-concepts-20040210/ RDF
Concepts and Abstract Syntax
<DanC> +1 "this is an RDF graph as defined in RDF concepts"
Mark: would this include Collection and Container? e.g. those would
not be our problem?
... can add metadata saying "these are a Collection" and that's
enough?
Jeremy: I would personally not get exercised if you left it that way
... but you might not be able to say you are "RDF complete", only "RDF
compatible"
... I don't think you can create an RDF graph consisting of a triple
whose subject and object are the same bnode
... that's OK; I'm not worried about that case
... fine goal to have something that is accessible to HTML authors,
this theoretical problem is less important
... there is a class of esoteric problems
... the one that worries me the most has to do with language tagging,
XML Literals and plain literals
... I intend to send an e-mail about this
Ben: would like an action to determine what parts of RDF cannot be
expressed in this XHTML syntax
Mark: we can add more attributes; e.g. a nodeID
... some readers will wonder what this is for and may never use it
DanC: an explicit algorithm from the XHTML2 syntax to RDF graphs
is preferable, to me, to an explanation of which bits are missing
... I want to see a positive mapping from this syntax to an RDF
Graph; this is more useful than identifying what is missing
Steven: we have such a mapping already
... we [HTML WG] agreed this morning to include this mapping as an
appendix to the XHTML 2 specification
Ralph: would this be a normative mapping?
Steven: yes, it is basically a GRDDL filter
Mark: this GRDDL thing is not quite finished
... we did a demo in Cannes but it needs to be updated
ACTION Mark: send pointer to GRDDL mapping for XHTML2->RDF
Jeremy: interested in Mark's comment that a nodeID attribute could be
added
... as this will only be confusing to an HTML author, am surprised
that Mark is so willing to add it
Mark: we haven't worked out what an ordinary ID on <meta> and <link>
really means
... there is an opportunity there to express something different
... we get anonymous nodes now with nested <link>
... I don't think that if one of these is named with an ID, does it
remain 'anonymous'? Is this up to the RDF processor whether to throw
away an ID?
DanC: anonymous nodes and things with ids are disjoint in the RDF spec
Mark: there's a difference between an ID on, say, a <div> for purposes
of HTML referencing and sticking an ID on a <link> or a <meta>
... <link> and <meta> make triples explicitly
... do we have to handle reification also or can we get away without
mentioning it?
Jeremy: one could discourage such a practice
<DanC> yes, reification is a dirty word. 1/2 ;-)
Ralph: keep ID in an XHTML document to refer _only_ to bits of (e.g.)
XML infosets and not to refer to abstract concepts. Don't reuse ID
to mean two different things within an HTML document.
<dom> [+1 to Ralph's point]
Mark: it seems a common use case to use ID to refer to the source of a
quote
Jeremy: referring to something in the body of a document is fine
... the are other mechanisms to refer to the subject
Mark: in some situations we work backwards; noting what an author
_could_ write and decide what it might mean
<Steven> "he is lost" or "he has lost"?
Mark: so if we have something that is the object of a triple, it could
mean several things
Jeremy: IDs in HTML documents refer to bits of documents
... some such triples might not be useful in an RDF sense
<dom> [10]http://www.ietf.org/rfc/rfc2854.txt says "For documents
labeled as text/html, the fragment identifier designates the
correspondingly named element"
<Steven> THis is most definitely not text/html, by the way
<dom> of course, what the semantics of an HTML element is not
well-defined afaik (e.g. HTML "infoset", ...)
<Steven> it is application/xhtml+xml
<dom> true
<dom> [11]http://www.ietf.org/rfc/rfc3236.txt has "For documents
described with the application/xhtml+xml media type, fragment
identifiers share the same syntax and semantics with other XML
documents, see [XMLMIME], section 5."
<dom> and XMLMIME says it's not defined
Mark: does reification fall into the same category as not being able
to express a bnode as the subject and object of a triple?
Jeremy: RDF Core did consider dropping reification
DanC: and I lost
Jeremy: I wouldn't expect a lot of pushback if XHTML2 does not support
reification
Mark: is the general thinking that OWL handles this?
<DanC> no, OWL doesn't provide anything that takes the place of
reification, FYI
Jeremy: even without explicit support for reification, if you can
construct an arbitrary RDF graph you can construct the triples that
reification needs
Mark: we will need to be able to make statements about other
statements
... so we may need the ID thing
Jeremy: reification does this not by referring to the triple but by
referring to the subject, predicate, and object of the triple
Ben: I think we will need to discuss [reification] more, after
Jeremy's email
Jeremy: I will try to show how to express reification in the XHTML2
syntax
ACTION JJC: comment on literals in section 19, 20 of XHTML 2
metainformation model
Ralph: [I wanted to mention PICS as a use case for making statements
about statements]
Ben: reviewing the requirements document produced by this TF
last year ...
[12]http://www.w3.org/2003/03/rdf-in-xml.html
[[
RDF in XHTML
Task Force Document 27 May 2003
]]
Ben: note that some requirements identified in that May document are
not addressed by this XHTML2 proposal
... e.g. not requiring RDF syntax to be rewritten
DanC: there may not be any special support for reification, but it may
still be possible
... specifically, I haven't seen anything done in FOAF that could not
be expressed in this syntax
Ben: about the requirement not to have to reformat RDF/XML?
DanC: I don't share that position, however it is reasonable for us to
say that the world has moved on
Ralph: I agree that what we have learned in the past 18 months can
permit us to re-evaluate requirements
Ben: relationship of Semantic Web to current Web
... the semantic web and the "clickable web" appear to be separate
... the links you click on a Web page are completely separate from RDF
triples
... there appears to be no way to link these two concepts together
... my position is that we should do more to link these two together
Mark: I agree, we've not done much to connect these
... e.g. <link rel='stylesheet'> doesn't actually require that there
be a stylesheet there and that a browser use it
... can't see changes to things as fundamental as <a> for anchor
... but to say that there is actually a predicate that gives a
specific type of relationship between two documents [that are
hyperlinked] does make sense to me
DanC: note keeping visible/navigable aspects of documents in sync
with formal knowledge is important to me, in practice, and motivated
the GRDDL design.
DanC: I came to the conclusion that sometimes <a href=> means 'this
is my brother', sometimes it means 'this is a document'
... seems a big challenge to address all of this in HTML space
... would like to see progress, but wouldn't suprise me if it turned
out to be hard
<mimasa> (just FYI: if anyone still wants to embed RDF/XML directly
into XHTML 2 and wants to validate,
[13]http://www.w3.org/2000/07/8378/schemas/nrl/xhtml2-rdf.nrl should
do the job)
Mark: role attribute brings in additional information from a known
taxonomy
... [role] gives scope for describing the relationship between two
documents as a known type
... e.g. a terms-and-conditions link might be in the header of a
document
... but the browser knows to render it at the foot because of its
particular role
<DanC> (does xhtml2-rdf.nrl say whether RDF in <blockquote> is
asserted or quoted, mimasa?)
<mimasa> (DanC, no, just validate XHTML 2 and RDF/XML concurrently)
Ben: so role might be useful to connect the semantics of the link with
the user-visible impact?
Mark: yes there are many applications of this, e.g. accessibility
Ben: I worry about existing links and approaches that could be
leveraged
... perhaps we can discuss this on the mailing list over the
next few days
David: note that RDF "metadata" may not always exist in XML
documents and that navigation into and out of "RDF Space" seems
required to implement a fully semantic Web.
... there are aspects of creating the Semantic Web that we cannot
address solely within the HTML WG
... but it is critical that we provide a way to link from HTML into
the Semantic Web
... so we need a syntax for embedding the semantics
... we need XHTML 2.0 to be able to link into the Semantic Web; it
would be catastrophic to miss this opportunity in XHTML 2
Mark: role does seem to handle this. Creative Commons would be another
use case
ACTION BenA: provide some Creative Commons use cases
Steven: Mark is suggesting <a href="..." role="..."> ?
Mark: yeah
<RichS> yes
Ben: aim for a next meeting in approximately 2 weeks
... will use mail to schedule
Summary of Action Items
ACTION: Mark to send pointer to GRDDL mapping for XHTML2->RDF
-- recorded in http://www.w3.org/2004/09/07-rdfhtml-irc#T13-31-54
ACTION: JJC to comment on literals in section 19, 20 of XHTML 2
metainformation model
-- recorded in http://www.w3.org/2004/09/07-rdfhtml-irc#T13-48-54-3
ACTION: BenA to provide some Creative Commons use cases [3]
-- recorded in http://www.w3.org/2004/09/07-rdfhtml-irc#T14-03-44
_________________________________________________________________
Minutes formatted by David Booth's [14]scribe.perl 1.90 ([15]CVS
log)
$Date: 2004/08/10 15:51:28 $
References
1. http://www.w3.org/
2. http://lists.w3.org/Archives/Public/www-archive/2004Sep/att-0010/telecon-2.html
3. http://www.w3.org/2004/09/07-rdfhtml-irc
6. http://lists.w3.org/Archives/Public/public-swbp-wg/2004Aug/0020.html
7. http://lists.w3.org/Archives/Public/www-archive/2004Sep/att-0010/telecon-2.html
8. http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2004Aug/0026.html
9. http://www.w3.org/TR/2004/REC-rdf-concepts-20040210/
10. http://www.ietf.org/rfc/rfc2854.txt
11. http://www.ietf.org/rfc/rfc3236.txt
12. http://www.w3.org/2003/03/rdf-in-xml.html
13. http://www.w3.org/2000/07/8378/schemas/nrl/xhtml2-rdf.nrl
14. http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribe.perl
15. http://dev.w3.org/cvsweb/2002/scribe/scribe.perl
Received on Tuesday, 7 September 2004 15:29:00 UTC