[uf-discuss] Citation: display-first?

[Bruce D'Arcus]

On 3/28/06, Michael McCracken <michael.mccracken at gmail.com> wrote:

...

> Basically, what I'm wondering is: if I'm marking up a citation, why
> does it matter if I sometimes need to do something like this:
>
> <a class="title url"
> href="http://www.library.yale.edu/scilib/modmodexplain.html">eprint
> Moderator Model</a>
> <span class="author vcard">
> <a href="http://pantheon.yale.edu/~dstern/dsbio.html" class="url
> fn">David Stern</a>
> </span>
>
> and sometimes this:
>
> <a class="title url"
> href="http://www.library.yale.edu/scilib/modmodexplain.html">eprint
> Moderator Model</a>
> <span class="author vcard">
> <a href="http://pantheon.yale.edu/~dstern/dsbio.html" class="url
> fn">David Stern</a>
> </span>

I've not yet had my morning coffee, but aren't the two above the same?

Also, what is it about the above that you think is different than the
dominant thrust of the existing discussion?

> Because allowing that seems more in step with the humans-first
> microformats methodology - we don't need to define a data format that
> can represent everything, we *do* need to give people a way to mark up
> citations they already produce in a way that provides more structure.

I think it's critical to get the basic structure of the modeling
right, and that will achieve what you ask for above.

The wrong approach is to just use a series of flat key-values, and to
borrow those from BibTeX.

The right approach is to define a basic set of relations, and a list
of properties.

Alf Eaton and I were discussing this just yesterday, in fact, and the
example he sent me (hope he doesn't mind if I post it here!) is close
to what I'd advocate:

<div class="bibliography" id="x-JMIR-v7i4e49-bibliography">
    <ol class="reference-list">
	<li class="citation reference book chapter" id="x-JMIR-v7i4e49-ref15">
		<a class="uri" href="urn:isbn:45346327"/>
		<span class="title">Advanced interactive video design: new
techniques and applications</span>.
		<span class="author name">Duppa N</span> and
		<span class="author name">Anderson K</span>.
		<span class="container">
			<span class="year">1988</span>.
			<span class="title">Book Title</span>
			<span class="editors">
				<span class="editor">Jane Doe</span> and
				<span class="editor">Simon Smith</span> (Eds)
			</span>
			<span class="publisher">
				<span class="location">New York</span>
				<span class="name">ABC Books</span>
			</span>
		</span>
		<span class="pages">
			<span class="page-start">33</span>-
			<span class="page-end">56</span>
		</span>
	</li>
    </ol>
</div>

My only complaint is that the "book" class really out to be on the
"container" span. I also think there are some other details to
address.  But I like it, and I think it's critical that the MF
recognize that bibliographic metadata is not flat, and that BibTeX
fields like "journal" are thus too limiting.

> Does anyone think I'm way off base here? I've started a use-case
> section on citation-brainstorming, and added my personal axe to grind
> - I'm interested to see other takes on what specifically a citation
> microformat would be used for.

To me your first use case is the most compelling. I think RSS/RDF
already does a better job at handling syndication than can be
reasonably achieved with microformats.

An aside:

I am on the OpenDocument Technical Committee. We've recently created a
metadata subcommittee, whose task is to add enhanced metadata support
to the file format.

It is highly likely we will be using RDF in some form. So for me, an
important requirement of any microformat is that it be easy to
transform to RDF (and back).

Bruce