W3C home > Mailing lists > Public > w3c-wai-hc@w3.org > July to September 1997

to classify dictionaries and stylesheets, view them as relations

From: Al Gilman <asgilman@access.digex.net>
Date: Fri, 19 Sep 1997 14:21:01 -0400 (EDT)
Message-Id: <199709191821.OAA12927@access2.digex.net>
To: w3c-wai-hc@w3.org (HC team)
to follow up on what Jason White said: 

> 2. I would like clarification of the proposal to reserve link
> types for accessibility-related resources. If the plan is to
> establish a key word which, if present, declares that the
> linked resource is intended for purposes of accessibility (for
> example an audio version of a document; perhaps in talking book
> format) then I think the approach is meritorious. However,
> certain types of resource that we have been discussing are not
> exclusively related to accessibility. For example, an
> abbreviation dictionary is important to speech output, but it
> may also be used by a spelling checker, or as a mechanism by
> which a link can be associated with each abbreviation, which,
> if activated, displays its expansion (this might be desirable
> in some educational settings, though the spelling checker
> illustration is a more commonplace application). Thus, care
> should be taken in deciding which link types should be
> associated with an "accessibility" key-word. As Daniel has
> suggested, CLASS may be a better location for the declaration
> of the type of dictionary (rel="dictionary"
> class="abbreviation").

The situation for dictionaries is a lot like the situation for
stylesheets.

For the dictionary itself, universally across all applications of
that dictionary, one wants to summarize the capabilities offered
by that dictionary.  This is a special case of describing the
functional class of a relation.  A dictionary is a relation in
that it associates text strings, the spellings of terms, with
other characteristics of those terms such as part of speech,
pronunciation, typical examples of use, idioms that use this
term, etc.  To summarize or characterize a dictionary, one wants
to communicate the following information items:

	domain of discourse where these terms are used:
		-- language and optionally professional discipline

	caracteristics enumerated in the dictionary:
		-- functional classes such as
			graph (spelling)
			part of speech
			date of earliest know use
			etymology
			pronunciation
			definition
			usage examples
			etc.

	encodings of characteristics
			language : ISO xxx extended per ...
			graph : charset as per HTTP
			pronunciation : code, e.g. IPA

This characterization constitutes the "pre-sale disclosure" information
that a potential user of this resource would want to have before
deciding to "buy" i.e. invest FTP or HTTP effort to get that resource.

The author of an HTML document may want to do two things:

	1: emphasize that a remote resource such as a dictionary
	is potentially value added to the processing of the
	current document.  This is the kind of information that
	I think the LINK is for.

	The reference to the dictionary would do well to characterize
	the capabilities of the dictionary [in the reference] if 
	the remote dictionary resource will be competing with local
	resources and perhaps other remote resources for net bandwidth
	and local compute cycles in the user's browse session.

	Or else the characterization is obtained by some lightweight
	form of metadata query such as RDF (PICS) or the HTTP HEAD request.

	2: steer the application of alternative resources when
	there is more than one resource with applicable capabilities.
	That is why one may wish to class a term 

	<SPAN CLASS=JARGON DOMAIN=IEEE/MTT>millimeter-wave</SPAN>

	as a technical term and therby force interpretation of that term
	through a specialist dictionary if the current document is
	a technical document that wishes to use jargon terms in their
	technical senses.

	What the HTML author does in this case is to narrow the 
	domain of application of the available dictionaries by giving
	preference rules to resolve application overlaps.  This
	can be done as a decision rule with dictionary selection as
	an output, or it can be done by projecting the capability
	type of a remote resource down to a sub-relation which is
	the slice that the current document wants to apply.  This
	creates a virtual resource which is a projection or slice of 
	the remote as-archived resource.

	Items in the document, whether text-level or block-level,
	can be bound to specific dictionaries by referencing the
	ID of a LINK element or an URL for the dictionary.  They
	can also be associated functionally by being associated
	with CLASS values and the dictionary selection controlled
	as a function of CLASS.  Or by an explicit reference in
	the other direction from a dictionary LINK to a 
	TARGET=id .

Let me strengthen my statement about dictionaries and stylesheets.
At the level of abstraction discussed just above, the theory of
relation classes and types fits both dictionaries and stylesheets
in the same way.

A stylesheet is a relation.  The User Interface is a collection
of resources.  The stylesheet binds content classes to realizations
which require those resources.  The MEDIA indication for a stylesheet
is an attempt to describe the general type of resources which that
stylesheet requires, as part of advertising the capability offered
by the stylesheet.

--
Al
Received on Friday, 19 September 1997 14:21:04 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 20:35:00 UTC