[uf-discuss] [citation]: Brian's outstanding issues 2:

[Bruce D'Arcus]

On 9/25/06, Ross Singer <ross.singer at library.gatech.edu> wrote:
> On 9/25/06, Michael McCracken <michael.mccracken at gmail.com> wrote:
>
> > The option of just ignoring types altogether - not including a type
> > property at all - is certainly possible - it would make human-reading
> > and publishing easier but automatic parsing somewhat harder. This
> > might be a worthwhile tradeoff.
> >
> I feel this is a very short-sighted decision, if it's the route hCite
> goes...  You'd never be able to link to an appropriate copy (because
> you wouldn't be able to determine with any semblance of confidence
> what an item actually is) and I'm therefore not sure what the point of
> this is.

I'm sort of agnostic about this (type or no type), but I'm not so sure
it follows that one must know a type of a resource to be able to find
a copy. Seems to me a good identifier (from which one can infer type)
is far more important.

> I guess the way I look at it is this: the entire point of formatting a
> citation in a standardized way is so that another scholar can then go
> in and know how to find the item again to follow the first person's
> research.  If the second scholar's /browser/ can do this work, the way
> that it looks on the screen is rather insignificant (much to the
> chagrin on the MLA and the APA, but they'll probably be happier in the
> long run).

Yes, but again: I don't think type is necessary for resolution. If I
have a URL, that's enough. If I have a DOI or an ISBN represented as a
URI, then also enough.

And when you deal with archival documents and such that I sometimes
deal with, they typically aren't web resolvable; at all.

> I've gone on record repeatedly that I don't care if hCite looks like
> OpenURL as long as it's easy to make something remotely OpenURL from
> it, but this is a fairly vital part of OpenURL... the very basic
> notion of knowing what something is.  I would recommend that we at
> least use the basic the journal (journal, article, issue, proceeding,
> conference, preprint), book (bookitem, book, proceeding, conference,
> report), dissertation (or thesis), or patent that are currently
> defined under the San Antonio profile in OpenURL

That would be a reasonable option, though I'd also suggest a more
generic "document" fallback (because real world citation practice just
doesn't fit pre-defined boxes). Also, *if* you have a container typed
as a "journal" then you also need a broader "periodical."

> (and it's actually
> trivial to add other formats if necessary -- yes, that's an open
> invitation to you, Bruce ;)).

Only so many hours in a day Ross ;-)

You can get a sense of what I consider important by looking at my RDF schema:

<http://purl.org/net/biblio>

Am currently revising it.

Worth noting that the classes have a hierarchy. So a Book is a
subclass of Document, and so forth. It makes the typing more robust.

Bruce