[uf-discuss] Citation format straw proposal on the wiki
bdarcus.lists at gmail.com
Wed Mar 29 04:10:32 PST 2006
On 3/28/06, Ross Singer <ross.singer at library.gatech.edu> wrote:
> On 3/28/06, Bruce D'Arcus <bdarcus.lists at gmail.com> wrote:
> > I have to disagree on the usefulness of the OpenURL stuff in this context.
> Can you explain this? w/r/t HTML, I find OpenURL the /most/ useful in this
> context, with this context being web content and OpenURL being a means to
> link a citation to an appropriate copy/service.
> In fact, I think if you use the 80/20 rule, your majority of users would be
> /much/ happier finding fulltext for a given citation than the ability to
> effectively load it into their citation manager.
Not aimed at you in particular Ross, but I really hate it when people
trot out the 80/20 rule, whose subtext is always about placing the
argument of others in the 20 category. And usually when people do it
in the context of citations, it is people who come from the hard
sciences telling the humanities people to be quiet (usually in the
form of "BibTeX is great, why use anything else?").
So if we do want to talk about 80/20 here, we need to clarify: whose
80/20? Do we only create a MF that works for geeks who code their own
HTML? Or do we consider a more inclusive approach that would be
appropriate for a broader range of users?
I'm not dismimssing OpenURL out of hand. Indeed, I added the "in this
context" qualifier, and certainly one could include OpenURL's (and
DOI's) within a MF. But I do reject the notion that the existing
OpenURL journal article schema, for example, provides a good model to
design a more general microformat. It's just flat key/value pairs,
which does not scale.
To me a test of an 80/20 format is can a user/developer reliably and
consistently encode the following:
1. articles (not just journal articles, but also for other periodicals)
2. speeches and other presentatiions (like a conference paper)
The trick is to avoid genre-specific property names like "jtitle" or
"conference-title" and exploit the nested possibilities of HTML and
the fact that one can include more than one class attribute.
But this does get us back to use cases and requirements. By your
logic, we might not bother with citation text at all. For me, though,
I want to be able to extract that content in addition to view it. We
could probably cover both needs though.
Aside: I typically send my publishers XHTML (generated from DocBook
and RDF source), so my full citations and bibliographic entries are
encoded in a microformat at that point. But there I have to conform to
precise publisher requirements about citation style.
More information about the microformats-discuss