W3C

RDF-in-XHTML Task Force

1 Aug 2006

Agenda

See also: IRC log, previous 2006-07-18

Attendees

Present
Elias Torres, Steven Pemberton, Ben Adida, Ralph Swick, Mark Birbeck
Regrets
Chair
Ben
Scribe
Ralph

Contents


Action Items Review

ACTION: [DONE] Ben fix content negotiation on his current personal Web hosting site to make http://ben.adida.net/card (w/o ".html") work again [recorded in http://www.w3.org/2006/07/18-HTMLTF-minutes.html#action07]

ACTION: Ben update the issues list [recorded in http://www.w3.org/2006/07/18-HTMLTF-minutes.html#action01] [CONTINUES]

ACTION: Ben start separate mail threads on remaining discussion topics [recorded in http://www.w3.org/2005/12/06-swbp-minutes#action04] [CONTINUES]

Ben: I need to reconcile those two and figure out what we're going to resolve this Fall

ACTION: [DONE] Ben to draft full response to Bjoern's 2004 email [recorded in http://www.w3.org/2006/01/24-swbp-minutes.html#action03]

Ben: just about to press the 'send' button
... Bjoern's resignation from validator work will unfortunately make this message appear to be something of a response to Bjoern's mail to W3C
... I have a sentence referring to Bjoern's "resignation" message

Ralph: strike that bit

-> Ben's reply to Bjoern's 18-month-old message

ACTION: Ben write a prototype hGRDDL profile for XHTML 1 [recorded in http://www.w3.org/2006/05/08-htmltf-minutes.html#action06] [CONTINUES]

Ben: Harry Halpin and Dan Connolly have asked me as Creative Commons representative to join the GRDDL WG and I have agreed to do so
... so I'm hoping that the hGRDDL work will be done jointly with GRDDL WG and become part of the GRDDL spec
... so it's pretty high priority for me to get an hGRDDL profile completed
... including a transform
... proposal so far is to have a profile URI that specifies a transform whose output is HTML
... the transform will not be specifically tied to RDFa

Elias: people were asking DanC whether GRDDL could be used to output other content-types than RDF
... Dan asked for official requirements to be submitted

<EliasT> ... http://chatlogs.planetrdf.com/swig/2006-07-31.html#T20-54-48

Ben: were people asking about other RDF serializations such as N3?

Elias: yes

Ben: GRDDL WG may start in the next couple of weeks

ACTION: Ben make sure RDFA bookmarklet runs locally [recorded in http://www.w3.org/2006/05/15-htmltf-minutes.html#action08] [CONTINUES]

ACTION: once Steven sends editors' draft of XHTML2, all TF members take a look and comment on showstopper issues only [recorded in http://www.w3.org/2006/02/06-htmltf-minutes.html#action01] [CONTINUES]

XHTML2: New Working Draft

-> XHTML 2.0 26 July Working Draft

<Steven> http://www.w3.org/TR/2006/WD-xhtml2-20060726/mod-metaAttributes.html#sec_24.2.

Steven: section 24.2 defines the actual relationship with RDF
... XHTML2 only comes in a "strict" version
... there is a separate specification for frames with a different media type
... all the "transitional" stuff is gone, though some has been moved into the core

Ralph: I'm impressed that all of RDFa fits on half a screen!
... I've only skimmed it and want to compare it thoroughly with our other RDFa detals

Steven: the XHTML2 specification document is not intended to contain tutorial material; tutorials are expected to be in separate documents
... in the case of XHTML4 spec it was hard to identify where normative specification ended and tutorial examples started

Ben: do CURIEs appear in the current draft?

Steven: no, we're using QNAMEs in content still as we haven't resolved the CURIE question
... the 26 July XHTML2 WD is not Last Call but we think we're very close to publishing a Last Call draft

Elias: does role appear in this draft?
... and does it match what we've decided in RDFa discussion?

Steven: role is in there but not mapped to RDF

Role attribute status

<benadida> primer: http://www.w3.org/TR/xhtml-rdfa-primer/

Ben: in the 16 May draft of the RDFa primer we use role as an abbreviation for rdf:type

Steven: I am very much in support of that approach

Mark: in long discussion history prior to the 16 May primer draft, role was html:role
... I feel that role needs to be a separate predicate
... we're really talking now about deciding whether both xhtml:role and rdf:type triples should be generated
... I think generating two triples is wrong
... as I've said in the past, I think class should be used for rdf:type
... unfortunately we haven't been able to get a clear opinion from the Accessibility folk
... I think if someone chooses to add the equivalence 'xhtml:role owl:sameAs rdf:type' to their own database that's fine but we shouldn't add it to the spec.

Elias: the XHTML spec will decide the definitions as XHTML relates to RDF and we should be consistent
... we should look at the impact on prior uses of HTML
... the meaning of role, though new to HTML, is not really like 'type'

Ralph: we have a new W3C staff person in the WAI area, Michael Cooper
... I would be willing to raise the question with Michael and see if we might get a formal opinion from the Accessibility folk

Elias: why do we need this shorthand?

Mark: when folks began to think about RDF in the HTML context, some people started using role and some people started using class

Elias: as soon as you need to specify two types you have to use rdf:type anyway
... this happens frequently in RDF/XML; there are documents with lots of rdf:type predicates
... converting FOAF is sort of a simple case
... what other kinds of documents need this shorthand?
... FOAF has been around a while and users have become accustomed to how it's used in RDF/XML
... I don't think people often check the type of a predicate; they "know" what it is
... are there examples other than FOAF where multiple types are clearly needed?

Mark: one question is whether the ontologies could have been designed better
... but one answer is that the design of the ontology is not up to us and people just want to be able to use existing ontologies easily

Elias: Henry Storey's ATOM ontology is very full and very descriptive
... in general, an ontology can describe the domain and range of its predicates
... people who want to know the classes of subjects and objects can load the ontology and make the appropriate inferences
... it's straightforward to deduce the class of a subject
... so how much detail really needs to be written into an HTML document?
... the important discussion is really whether we need a shorthand
... nothing prevents an HTML author from using rdf:type explicitly
... do we have more examples that show a shorthand is really required?

Ben: Elias is coming from the RDF point of view; we'd need to consider this from the point of view of an HTML author as well
... e.g. in microformats, people declare Events, etc.
... they anchor their type declarations using class=
... if we make RDFa more complicated than this then we will lose part of the community

Elias: most people see subjects and objects as having just one type

Mark: another approach would have been to do a full review of all the features in HTML and see what we could use them to specify
... in some sense the microformat people have done this

Elias: if we do go with a shortcut, I think class is going to be better, more elegant

Mark: it would be wrong not to use class for _something_
... and others have chosen to use class for type

Steven: I have a lot of sympathy for using class=
... but with my chair hat on I see a big hassle because class is very widely used by a community who do not necessarily subscribe to our goals
... e.g. CSS
... if we start assigning meaning to class values - especially as it likely changes the current meaning, given that class values currently do not have a namespace -- I forsee a lot of trouble
... we will have triples produced that were not intended by the authors
... if we use role we won't have to deal with this legacy

Elias: currently XHTML2 doesn't say how role relates to RDF

Steven: only because we haven't decided exactly what the relationship is
... we agree that xhtml2:role should have _some_ relationship to RDF
... it's in the draft now at the request of the WAI community
... WAI and Device Independence/Mobile are the two communities who are interested in using role
... and both are interested in the metadata [triples] produced

Mark: there are best practice statements that some URI points to an RDF schema

Elias: but type is different from role

Mark: that is my view
... role='toolbar' is not saying something _is_ a toolbar but only that it plays the function of a toolbar

Mark: I think Ben's solution for rel= and rev= works quite nicely
... unprefixed values come from a specified list
... prefixed values come from the corresponding namespace
... so people won't be making globally meaningful statements without intending to do so
... but many writers are suggesting ways to add more semantics to HTML markup and they're recommending the use of class=

Steven: changing the syntactic value of class

Mark: changing the syntax of values doesn't change the meaning

Steven: I foresee a huge controversy with other users of class

Mark: but if we try to persuade users of microformats to adopt RDFa and say nothing about the class attribute then we're setting ourselves up for failure

Elias: we do have to think about adoption
... just because we can define a brand new XHTML container format doesn't mean we'll convince people to use it

Ben: we'll need to continue this conversation; we're seeing here the many faces of the community

ACTION: Ben check that the issues list summarizes all the viewpoints on role and class attributes [recorded in http://www.w3.org/2006/08/01-htmltf-minutes.html#action08]

Steven: including that role is a shorthand for a predicate

Mark: I'm proposing that we take as a given that role does generate an xhtml:role triple and the question is whether it also generates another triple

QNames in Attributes

Ben: I sent personal mail to Ian Davis
... as he's done work in that area
... I believe that XML purists deplore QNames in attributes but have come to accept it

Elias: would that mean we no longer need CURIEs?

Ben: no, it's just a matter of whether it has become an acceptable practice to use QNames in attribute values

Elias: QNames in href?

Ben: not allowed. Only possible with CURIE approach

Elias: I'm not sure I'd even want QNames/CURIEs in href
... I'm concerned about what would happen with [current] browsers

Ben: yes, that's another discussion

<MarkB_> http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2006Jun/0009.html

Next meeting

Ben: can everyone make next Tuesday?

<EliasT> I can.

Steven: yes, but not the following 3 weeks

<Ralph> I expect to be able to participate on 8 Aug

Next Meeting: 8 Aug

XHTML1.1: status of the module and what we can do to begin validating

<MarkB_> http://skimstone.x-port.net/node/213

Mark: there's been some progress on a schema that works with a validator

<EliasT> is it this: http://schneegans.de/sv/

Ben: there was an announcement of a validator that supported namespaces

Mark: yes, schneegans was using modularization schemas. I should contact them.

Ben: I'd like to see a concise document that points to the current work and a validator that does work, along with renderings in current browsers

ACTION: Mark contact schneegans.de folk about connecting with our work [recorded in http://www.w3.org/2006/08/01-htmltf-minutes.html#action09]

<Steven_> http://relaxed.vse.cz/Wis the one I meant

<MarkB_> thanks

Elias: I'm not hearing a huge validation requirement right now

<EliasT> Ben was asking whether I had any specific XHTML 1.1 requirements. I am mostly interested in reducing the number of issues people have with RDFa. If we can solve one great. At the moment, I don't have any hard requirements.

Summary of Action Items

[NEW] ACTION: Ben check that the issues list summarizes all the viewpoints on role and class attributes [recorded in http://www.w3.org/2006/08/01-htmltf-minutes.html#action08]
[NEW] ACTION: Mark contact schneegans.de folk about connecting with our work [recorded in http://www.w3.org/2006/08/01-htmltf-minutes.html#action09]

[PENDING] ACTION: Ben make sure RDFA bookmarklet runs locally [recorded in http://www.w3.org/2006/05/15-htmltf-minutes.html#action08]
[PENDING] ACTION: Ben start separate mail threads on remaining discussion topics [recorded in http://www.w3.org/2005/12/06-swbp-minutes#action04]
[PENDING] ACTION: Ben update the issues list [recorded in http://www.w3.org/2006/07/18-HTMLTF-minutes.html#action01]
[PENDING] ACTION: Ben write a prototype hGRDDL profile for XHTML 1 [recorded in http://www.w3.org/2006/05/08-htmltf-minutes.html#action06]
[PENDING] ACTION: once Steven sends editors' draft of XHTML2, all TF members take a look and comment on showstopper issues only [recorded in http://www.w3.org/2006/02/06-htmltf-minutes.html#action01]

[DONE] ACTION: Ben fix content negotiation on his current personal Web hosting site to make http://ben.adida.net/card (w/o ".html") work again [recorded in http://www.w3.org/2006/07/18-HTMLTF-minutes.html#action07]
[DONE] ACTION: Ben to draft full response to Bjoern's 2004 email [recorded in http://www.w3.org/2006/01/24-swbp-minutes.html#action03]

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.127 (CVS log)
$Date: 2006/08/01 14:30:42 $