ZedAI telcon 20081124

From zedwiki

Jump to: navigation, search

Contents

Present

Regrets: Matt

Scribe

Boris

Agenda

Revisit suitability of Telcon Time (Kenny, Sam)

Action Item Review

  • Markus fix ZedAI main page, start a ZedAI Overview page
  • Markus to follow up the contact with the xhtml2 WG.
  • Markus/Per - discuss pros/cons of the XHTML2 modularization framework (started at ZedAI W3C Modularization and XHTML2.
  • Per: continue frontmatter content model discussion on list (ref: RNIB Book DTD)
  • Per: reiterate em+strong at list
  • Matt: send ToC example to list
  • Per: compare of Matts and Stephens ToCs
  • Gregory: post NewsML G2 info to Wiki, notify list.

Simple Profile Content Model Topic: toc

  • current proposal: use James's model with enhancements through metadata annotation 
<tocentry x2:property="bana:mainEntry">
    <label>INTRODUCTION</label>
    <tocentry x2:property="bana:subEntry">
          <label>Blank Is Beautiful: Three Decades of Erasing and Remaking the World</label>
          <pageref ref="p3">3</pageref>
    </tocentry>
    ... etc.
</tocentry>

Simple Profile Content Model Topic: frontmatter

  • question from James: do we really need front/body/rear matter?
  • proposal: base on RNIB DTD (match with Chicago Manual of Style)

NewsMLG2 (Kennys question on whats needed) 

Dear All,
 
NewsML-g2 can encapsulate the following types of information:
1. Metadata about the News.
2. News management metadata.
3. metadata about the packaging.
4. Information about the technical transfer.
5. news Content.
6. Concepts (worth-to-remember information)
7. Knowledge (a collection of concepts)
 
You can read more about each of these items at [1].
 
Which of these pieces of information should be captured in the Periodicals profile? 2,3 and 4 seem to be the obvious exclusion items, however, 3 might still be required.
 
Your thoughts?
 
[1] http://iptc.cms.apa.at/cms/site/index.html;jsessionid=alSz8rIEE938?channel=CH0123
 
Warm Regards,
kenny.

Minutes

Time of Meeting

To be moved one hour later: 14:00 UTC through end of December.

Action Items

Completed

  • Markus/Per - discuss pros/cons of the XHTML2 modularization framework (started at ZedAI W3C Modularization and XHTML2
  • Matt: send ToC example to list
  • Per: compare of Matts and Stephens ToCs
  • Per: reiterate em+strong at list
  • Gregory: post NewsML G2 info to Wiki, notify list.

Ongoing

  • Markus fix ZedAI main page, start a ZedAI Overview page (postponed, Markus is working on spec prose first.)
  • Per: continue frontmatter content model discussion on list (ref: RNIB Book DTD)
  • Markus to follow up the contact with the xhtml2 WG.

Gregory: WG voted to move ahead on all 1.1 modules, leaving them free to concentrate on XHTML2. The Device Independence WG is planning to use XHTML2 as base - like us, this group is concerned about and unsure of status. Will put Markus in contact so we can work together. XHTML2 WG will be discussing status and dates on Wednesday.

DAISY is now a member of XHTML2 WG with Markus as rep.

TOC Format

Current proposal: use James's model with enhancements through metadata annotation

As discussed on mailing list, this is not sufficient for BANA Braille.

Markus: our general strategy should be to cover 80% use cases with simple structure, use annotations to allow organizations to fill in and add detail as needed for rare cases.

For annoation - not sure which mechanism to use - role? property? RDF/A? Seems like there are several generations of solutions in the spec.

Gregory: RDF/A is going through. So is ARIA. Current metainformation module of XHTML2 will be changed in favor of RDF/A.

Could put aria roles onto <label> elements to show more detailed structure.

TOC items pointing into frontmatter and backmatter can be formatted differently. How to indicate this?

XML World migrating towards RDF/A - may be best solution for us. ARIA targeted towards AT; RDF/A is more abstract. In this case may want to use both.

All seem agreed to use some sort of annotations to allow content models to be kept simple, avoid needing new modules & profiles.

Gregory: Generally, use aria role for the parent object or wrapper. Eg for labeling entire TOC as a TOC. Inside, use aria:labeled-by to connect back to label of larger container. Predefined aria roles are pretty general, but we can make more specific roles through extension mechanism.

Is the use of annotations going to leave things too loose? Preferable to have everything be validatable. You can validate XHTML + RDF/A now - need more than just DTD validator, but these annotations are from a controlled (though extensible) vocabulary, so they can be as precise as XML elements.

TOC, or parts of it, could be automatically generated. But some organizations need the ability to replicate the print book, including any exceptions and inconsistencies.

Risk: different organizations may use optional elements and annotations in different way, and we lose consistency.

Disadvantage of annotations: May be more difficult to deal with attributes than elements in an XML editor. However, we are expecting that the annotations would only be needed by the more unusual use cases so this may be an acceptable tradeoff.

Gregory: WG is reexamining what basic types of lists there are in XHTML2 - might add a new list type that is useful here.

Simple Profile Content Model Topic: frontmatter

Will move forward with creating an example/recommendation here, and discuss next week.

NewsMLG2 (Kennys question on whats needed)

Kenny: The NewsML G2 DTD is not modular.

There are two schemas: core g2 architecture, and news item element.

There are specific vocabularies for various content types - eg sportsML. But these are not broken into modules and assembled in a driver like XHMTL2; they are separate monolithic schemas.

NewsItem element is the holder of news content. It allows ANY XML vocabulary inside it - typically content is XHTML, but NewsML schema is ultimately lax.

So it does not look like we can borrow or import elements from NewsML - we will instead need to create our own elements and bolt them onto XHTML. Perhaps we can refer back to NewsML grammars through role or RDF to show correspondence.

Some useful elements we may want to consider: headline, byline, dateline

NewsML schemas are wrapper languages, not specific like MathML. Associate NewsItem (content) elements with PackageItem, ConceptItem, KnowledgeItem which express connections to controlled vocabularies of concepts. A lot of effort going into things probably not useful for DAISY: news management and distribution, and few strong semantic types of the sort that we were looking for. Looks like target audience of these two specs are widely different.

Kenny was optimistic until he looked at the schema. There are multiple schemas, and several elements are duplicated across namespaces rather than reused. We would certainly not want to adopt the lax "ANY" schema, that allows absolutely any XML content.

Are there any semantics in news management that we might be interested in? We may be interested in some of the metainformation. News codes are robust information about the context of the target items. Gregory will put some pointers to useful bits onto the wiki.

Suggestion: Kenny to do a couple of samples with XHMTL + a couple of new elements like headline, byline, etc.

Is it worth looking at NITS? Ole: you might as well use XHMTL. A lot of work, but doesn't buy us that much.

Magazine DTD - will take a look at that as well.

Small News profile group (Gregory, Ole, Kenny, Markus) will meet Wednesday

Retrieved from "http://www.daisy.org/zw/ZedAI_telcon_20081124"
Personal tools