Open eBook(TM) Publication Structure 1.0

Sept 23, 1999 the OEB meeting in Gaithersburg, MD (USA) reported
on the Version 1.0 E-book XML application, released Sep 16, 1999.
     http://www.openebook.org

"One element of the Open eBook initiative is a specification for eBook file
and format structure based on HTML and XML, the languages used to format
information for Web sites. The goal of the specification is to quickly 
create a
critical mass of compelling content. A publisher will be able to format a 
title
once according to the specification and the content will be compatible with
a wide variety of reading devices. Agreeing on a common set of file 
specifications
will allow publishers to reach a large audience without separately 
reformatting
their titles for each machine. This specification is designed to be compatible
with the development plans of the major eBook efforts already underway."

Markku Hakkinen and George Kersher were on the design committee.
Mark indicated he was satisfied that it contained [some but
not all] of the accessibility features in strict HTML 4.0 DTD.
[They have emasculated the HTML 4.0 table model, and
have reverted to only table, row, td and th.]

Open eBook Publication Structure 1.0 defines the format that
content takes when it is converted from print to electronic form.

A particularly significant capability is a package DTD
that identifies what's in the book, and includes the Dublin Core
metadata as elements (rather than embedded as attribute
name="..." value="..." pairs of the HTML 4.0 meta element),
a manifest of all component files, and a spine that gives major
ones in their normal order in the book.

The major parts of the OEB package file are:

PACKAGE IDENTITY – a unique identifier for the OEB publication as a whole.
METADATA – Publication metadata (title, author, publisher, etc.).
MANIFEST – A list of files (documents, images, style sheets, etc.) that 
make up
     the publication. The manifest also includes fallback declarations for 
files of
     types not supported by this specification.
SPINE – An arrangement of documents providing a linear reading order.
TOURS – A set of alternate reading sequences through the publication, such
     as selective views for various reading purposes, reader expertise 
levels, etc.
GUIDE – A set of references to fundamental structural features of the 
publication,
     such as table of contents, foreword, bibliography, etc.

The Open E-Book view of markup is that the XML markup in any
e-book must be well-formed (all start tags properly closed,
and properly nested). They do not require validity. That
allows any arbitrary tags and attributes on start-tags to
occur in the markup.

Presumably semantics are associable with such arbitrary tags,
and at least one appropriate style description should be
provided for them. That doesn't necessarily preclude providing
accessible style description, though there is no such
requirement at this time. CSS-1 and some of CSS-2 are expected
to be supported.

Ben Trafford, one of the advocates speaking about the
current design of their DTD, said that he would recommend
using the extensive Text Encoding Initiative (TEI) DTD and its
modules (recently converted from SGML to XML) for any
e-book he would mark-up, and just cover them with a minimum
of OEB tagging.

This approach seems to me to allow substitution of major
material for which accessibility issues may not have been
addressed. For example, the scholarly criticism aspects of
TEI are not automatically accessible.

Identifying presentation styles, whether visual, audible,
or tactile, is a major task, particularly when contextual
differences are significant, for all contexts in which
tags can occur. Recursive tags desirably use recursive
styles, such as incremental indention or smaller font
pointsize for visual differentiation. When prefix
material is added, such as bulleting or enumeration of
lists, the recursive model has more challenges. Clearly
the number of recursive levels have practical limits in
any book. Making those level distinctions audible or
tactile may get lost unless aided by structural analysis
(and possibly queried by the reader) say of the reading
context found from the Document Object Model.

Any book may need multiple stylesheets, designed to the
particular book, one each for the different supported
output media. A publisher could use the house styles
and supply them with all its books, so that the style
needed by the user and rendering equipment can be linked
to for use.

WAI needs to remain aware of the work on open E-Books, to
assure that accessibility concerns in the next version are
addressed adequately.

Regards/Harvey Bingham