- From: Ralph R. Swick <swick@w3.org>
- Date: Fri, 29 Oct 2004 15:51:22 -0400
- To: public-rdf-in-xhtml-tf@w3.org
SemWeb BPD HTML Task Force
29 Oct 2004
[2]Agenda
See also: [3]IRC log
Attendees
Present
Ralph Swick, Jeremy Carroll, Dan Connolly, Ben Adida
Regrets
Mark Birbeck
Chair
Ben
Scribe
Ralph
Previous: 2004-10-22
[9]http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2004Oct/0024.html
Contents
* Topics
1. Action review
2. Jeremy's XSLT2 implementation of RDF/A
3. Jeremy's Comments on RDF/A
* Summary of Action Items
_________________________________________________________________
ACTION: Ben report back to HTML WG on our evaluation of RDF/A
Action review
ACTION BenA assesses impact of namespace and media type change on CC
BenA: talked with Mike, CTO of CC
... happy to have an XHTML 2 solution but we expect adoption to not be
immediate
... CC has to decide how to move to a notation that will be
XHTML2-compliant and that is going to at least render correctly in
current deployed browsers and ideally that would validate
... may adopt the current RDF/A syntax inside our recommended HTML
BenA: ... regarding literals, considering Jeremy's comments it appears
we are in fine shape
... we don't see a need for XML literals
... we do need plain literals
<DanC> (hmm... is there an XHTML2 validation service yet?)
JJC: do you care about having to duplicate content?
BenA: I am a big supporter of not having to duplicate content, yes
action complete
ACTION BenA determine CC requirement with respect to literals
-- complete (see above)
ACTION BenA to find someone to determine requirement with respect
to FOAF and literals
BenA: will be contacting Dan Brickley
action withdrawn, replaced with ACTION JJC below
ACTION JJC to check nodeID and bnode issues
-- completed
ACTION: JJC contact Dan Brickley to determine requirement
with respect to FOAF and literals
ACTION Steven and Mark to add about="" to example 6.1 (CC)
[10]http://www.formsplayer.com/notes/rdf-a.html appears to have last
been modified on 2004-09-22
-- continues
ACTION DanC: ask for contact from SHOE/DAML/OWL "how do I put this in
my web page?" community
-- continues
DanC: what is the time window for getting a SHOE/DAML/OWL person into
the discussion and up to speed?
BenA: HTML WG meets f2f week after next and hopes to go to Last Call
shortly thereafter
JJC: want to talk about complexity. This may hurt the HTML WG schedule
DanC: GRDDL does not depend on the HTML WG taking further action
Ralph: I fully expect this TF to come back to GRDDL discussions once
we've gotten our feedback to the HTML WG
Jeremy's XSLT2 implementation of RDF/A
[11]http://lists.w3.org/Archives/Public/www-archive/2004Oct/att-0052
JJC: I just updated the XSLT2 implementation and now believe it to be
a complete implementation of RDF/A except for the bnode() Xpointer
scheme
DanC: I suggest recruiting a reviewer, or at least an alpha tester
ACTION: Ben test JJC's implementation on CC cases
<jjc> (oh there's a bug to do with whitespace within XMLLiterals)
JJC: the key goal of the implementation was to have a clear mapping
from the document to the code
... the critical sections (4 and 5) correspond paragraph-by-paragraph
<DanC> (hmm... is it worth putting the "simple rules using XPath
expressions" in the RDF/A spec?)
JJC: there are two transformations from the doc to the final XSLT2
code
... it's meant to be obvious where to change the code if the spec is
changed
Jeremy's Comments on RDF/A
[12]http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2004Oct/0047.html
RDF/A review summary [Jeremy 2004-10-28]
JJC: the I18N WG raised a formal objection to the RDF Core's decision
on xml:lang in XML Literals
<DanC> (my instinct would be to get the RDF/A spec to the point where
the rules could be scraped from it, ala "ical RDF schema: derived from
the RFC"
[13]http://lists.w3.org/Archives/Public/www-rdf-calendar/2004Mar/0007.html)
JJC: RDF Core decided to simplify XML Literals so they were just like
other datatypes w/o an xml:lang
... using <span> to add xml:lang back when necessary
... if RDF/A goes to Last Call as currently written the I18N folk are
likely to again raise a strong objection
DanC: I don't care much about the details until I hear someone
actually using a feature
... RDF Core spent a year on the details of language tagging and I'm
not aware of anyone using it
... people are creating new properties
JJC: we have Jena users using xml:lang
BenA: tempted to say this is not in our scope
... as the same problem occurs in RDF/XML
JJC: within RDF/XML the authors are explicitly told that xml:lang is
not in scope and if they want it they must include it explicitly
... whereas with XHTML the expectation is that xml:lang is in scope
... e.g. language-dependent stylesheets would be expected to work
inside this XML markup
DanC: consider the W3C translation system as a use case
[14]http://www.w3.org/Consortium/Translation/
... I think Ivan Herman is the person responsible for the W3C
translations database
... think he maintains that by hand, not by automated scraping
... I18N WG might be able to point to use cases
... we could point out to the HTML WG the risks of this current
solution
JJC: I have a solution, which doesn't require the HTML author to do
anything more; the mapping (XSLT) adds the <span>
RESOLVED to point out to HTML WG the risk of I18N objections to the
RDF/A handling of xml:lang but that we feel a solution exists that
requires no additional work for HTML authors
JJC: regarding literals ...
... last week people were interested in having plain literals in
element content
[15]http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2004Oct/0037
<DanC> datatypes + in-line plain literals
JJC: my suggestion is to take the xpath text if there is other markup
present; i.e. ignore the markup
DanC: example would be more compelling if it used something other than
<meta>, as <meta> is already special
JJC: this would be easy to implement; didn't do it in order to stay
strictly to the written spec
... two ways to denote a plain literal in element content
... 1. special datatype 2. separate attribute
DanC: I prefer a separate attribute, as the type of the datatype
attribute is QName
Ralph: we could make a plain literal QName
JJC: in RDF plain literals have a lang tag so they're not typed
... prefer separate attribute approach
<DanC> (yes, "ugh" applies to almost all aspecs of datatypes in RDF.)
<jjc> .//text()
<jjc> takes the text() nodes of descendenents
<DanC> yes, let's ask them to replace the metainfo module with RDF/A
RESOLVED to propose to HTML WG a 'plain="true"' attribute for
expressing plain literals in element content
PROPOSE We find RDF/A a big step forward and encourage the HTML WG to
use it in place of the 22 July metainformation module. Our detailed
criticism of RDF/A is intended to helpfurther improve this work toward
the long-standing needs of the RDF deployment community to mix
semantic web data with HTML documents.
(consensus among the 4 of us)
<benadida> ACTION: BenA to write up a version of the above and send to
list
<DanC> Connolly 2nds that and trusts ben to stay in that
neightborhood.
JJC: there are quite a number of ways to determine subjects and
objects and the objects are particularly complicated
... it's very hard to keep track of this all
... my implementation does keep track in an elegant way that is hugely
inefficient
... the resulting complexity is too much for humans to understand and
sufficiently hard that programs will frequently be wrong as well
... getting rid of the the predicate inheritance rules would help some
... the subject rules amount to 8 different ways to determine a
subject
... this is a lot but not too bad
... in paragraph 4.4.3 there is a rule
... different treatment for link & meta elements than all other
elements
... most of the examples in RDF/A actually produce more triples than
the authors intend
... the current draft is good but not yet REC-quality
... it's not yet clear which rules need to be removed; this is a
complex language design issue
JJC: the essential idea of 4.4.3 introduces a huge complexity
JJC: but I also like 4.4.3 as well
RESOLVED: the current rules provide more options than needed and
produce more triples than intended. We offer to work with the HTML WG
to simplify the rules for determining resource-valued objects
Discussion during f2f?:
Ben: I could be available 8:30am-1:00pm EST on Monday
Summary of Action Items
ACTION: DanC to ask for contact from SHOE/DAML/OWL "how do I put
... this in my web page?" community
[NEW] ACTION: Ben report back to HTML WG on our evaluation of RDF/A
[NEW] ACTION: Ben test JJC's implementation on CC cases
[NEW] ACTION: BenA to write up a version of the resolution on RDF/A
... and send to list
[NEW] ACTION: JJC contact Dan Brickley to determine requirement
... with respect to FOAF and literals
_________________________________________________________________
Minutes formatted by David Booth's [17]scribe.perl 1.90 ([18]CVS log)
$Date: 2004/08/10 15:51:28 $
References
2. http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2004Oct/0048.html
3. http://www.w3.org/2004/10/29-swbp-irc
9. http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2004Oct/0024.html
10. http://www.formsplayer.com/notes/rdf-a.html
11. http://lists.w3.org/Archives/Public/www-archive/2004Oct/att-0052
12. http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2004Oct/0047.html
13. http://lists.w3.org/Archives/Public/www-rdf-calendar/2004Mar/0007.html
14. http://www.w3.org/Consortium/Translation/
15. http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2004Oct/0037
16. http://www.timeanddate.com/
17. http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribe.perl
18. http://dev.w3.org/cvsweb/2002/scribe/scribe.perl
Received on Friday, 29 October 2004 19:54:27 UTC