RE: A note about the definition of "structure"

This is a very interesting and thoughtful posting.   


Your comment:
" First of all, a definition have to explain what something _is_, not
what something _includes_."
is one we should look at across all of our definitions. 


You also make a good point (actually several) about hierarchy and
relationship.   We are schizophrenic about it the way it is currently both
in and alongside structure.  We need to fix that. 

But I'm not sure which was is correct.   With the web we can have links that
provide relationships that are more weblike than tree like.   Would you call
these hierarchical?   Think of a 5 pointed star with a line between all
points. 

That makes me think that relationships exist that are not hierarchical. 

We then have several choices 

OPTION 1:
SC "Relationships within the content can be programmatically determined."
with
Definition:   Relationships include (oops the include word) Hierarchical and
non-hierarchical relationships.


OPTION 2:
SC "Structure within the content can be programmatically determined."
with
"Structure is the full set of relationships existing between the
contents of a web page. Relationships can be horizontal, as between
two items in a list, or vertical, as between a header and the
paragraphs and subheadings associated with it or XXXXXXX as between related
sections that link to each other."



I had trouble however following your comments about separation of content
and presentation.  It is true that you use presentation to convey structure.
But if you use markup to indicate structure (e.g. headers) you have
separated structure from presentation.   The browser then determines how the
structure will be presented. 

Also please look closely at the guideline. 
"The Ensure that information, functionality, and structure are separable
from presentation."
It doesn't say that they can't be confounded - just that they must be
separable.  That is - you must be able to derive the information and
structure even if you can't view the presentation and must cause the
information to be presented in another fashion.  If the structure and
information cannot be extracted from or is not independently indicated then
it will be lost in the re-presentation. 

Is it clearer now what is intended?   Is there a better way to word this to
make it clearer? 


Gregg

 -- ------------------------------ 
Gregg C Vanderheiden Ph.D. 
Professor - Ind. Engr. & BioMed Engr.
Director - Trace R & D Center 
University of Wisconsin-Madison 

-----Original Message-----
From: w3c-wai-gl-request@w3.org [mailto:w3c-wai-gl-request@w3.org] On Behalf
Of Michele Diodati
Sent: Sunday, January 16, 2005 8:22 AM
To: w3c-wai-gl@w3.org
Subject: A note about the definition of "structure"


This is a little contribution, hoping it will be useful to make as
clear as possibile the meaning of the word "structure" in WCAG 2.0.

The first part of the current definition of "structure" says [1]:
"Structure includes both hierarchical structure of the content and
non-hierarchical relationships such as cross-references, or the
correspondence between header and data cells in a table."

I disagree with this definition for some reasons. 

First of all, a definition have to explain what something _is_, not
what something _includes_.

In the second place, a definition that includes in its explanatory
part the same word it is trying to explain ("structure includes both
hierarchical structure..."), isn't really explaining nothing.
Philosophically speaking, this is a kind of tautology.

Thirdly, in my opinion "not-hierarchical relationships" do not exist.
Every piece of content in a web page is part of a hierarchical
structure. A hierarchy is made of relationships, and relationships can
be horizontal or vertical. From a logical viewpoint, between a header
and the paragraphs associated with it, there is a vertical
relationship: in a certain sense the associated paragraphs are "sons"
of that header. On the contrary, between two cross-references within
the same web page there is a horizontal relationship: they are like
two students from two classes of the same level in two different
sections of the same school. Both students are at the same
hierarchical level and their common origin in the hierarchy is the
fact their two classes belong to the same school (examples could be
very numerous). There is an association between the two ends of a
cross-reference. This association has a semantic nature: leaving aside
semantics, only their horizontal relationship (association) remains.

So, I propose a re-wording of the first part of the definition of
"structure". Here is my proposal:
"Structure is the full set of relationships existing between the
contents of a web page. Relationships can be horizontal, as between
two items in a list, or vertical, as between a header and the
paragraphs and subheadings associated with it."

Morevover I propose to modify the wording of SC L1 for GL 1.3. It
actually says: "Structures and relationships within the content can be
programmatically determined."

If you agree that structure _is_ made of relationships, then it seems
to me a source of potential misunderstanding to use the formula
"structures and relationships". I would modify the wording in this
way: "Structure of the content can be programmatically determined."

Or else: "Relationships within the content can be programmatically
determined." (eliminating the word "structure").

Also, it seems to me a little problematic the same GL 1.3, which
actually says: " Ensure that information, functionality, and structure
are separable from presentation."

I think presentation is a kind of information in itself. Besides
presentation is the only way in which you can show contents and their
structure to the user. For example, a printed page shows visually,
through their sizes, the reciprocal relevance of the headers used in
the text. Without a presentation (any kind of presentation), no
content nor structure remain. Also disabling style sheets support in a
browser, a simple presentation remains, showing through sizes,
positions and emphasis horizontal and vertical relationships between
the contents.

Having said that, I would modify GL 1.3 in a way similar to this:
"Ensure that presentation is consistent with the structure of the
content".

According to me, presentation isn't really separable from structure:
it is rather the main way in which you can show the structure of the
content. This has nothing to do with the necessity to use style sheets
instead of (X)HTML code to determine the presentation in a web page.
You can obtain a complete separation between semantic code and
presentation code, without having a more accessible page. For example,
it is possible to use style sheets to format a more relevant header
with a font size smaller then the size used for a less relevant
header: the separation at level code is perfect, but this kind of
presentation is not consistent with the intrinsic structure of the
content. So it seems to me more important for accessibility that GL
1.3 states that presentation has to be consistent with the logical
structure of the content (because presentation is the only way to show
the logical structure of the content).

The separation between presentational code (CSS) and semantic code
(structural elements of XHTML, or XML elements) is possible, but it is
not equivalent to a real separation between presentation and
structure. Indeed, the goal of any presentation is to convey the
structure of the content to the users.

Ciao,
Michele

[1] http://www.w3.org/TR/2004/WD-WCAG20-20041119/#structuredef

----------------------------------
M i c h e l e    D i o d a t i
Via Pian due Torri 86 - 00146 Roma
Tel. 06 5503533 - Fax 06 233212132 
http://www.diodati.org
----------------------------------

Received on Monday, 17 January 2005 04:34:13 UTC