[openbiblio-dev] Feedback on my BibJSON record
kcoyle at kcoyle.net
Fri Feb 22 15:08:54 UTC 2013
It would really be great to have a list of BibJSON terms, and a clear
statement about the BibJSON format. If they exist, I can't find them.
There was a discussion about using JSON-LD, but the last thing I can
find is a blog post saying it is being considered . Looking now the
bibJSON.org page says:
We would like to use real world examples as far as possible - do you
require linked data functionality within BibJSON? If so, please contact
us and we can work through a BibJSON / JSON-LD example with you."
And just below it is:
"There is no fixed schema as yet, but as JSON-LD is supported it is
possible to reference any vocabulary via a namespace declaration and use
it with your key/value pairs where necessary."
So I would like to make a call for some clarity in the documentation,
and for documentation beyond the one page of examples. I hadn't seen
that JSON-LD had been accepted, and I have to surmise it from "but as
JSON-LD is supported...." which isn't ideal as documentation. I believe
that there are terms that are considered accepted terms for bibjson -- a
list of them should be made available, preferably in a
machine-actionable form as well as human-readable.
That said, if one needs to cover the library distinctions of title,
subtitle, and statement of responsibility, it would probably be best not
to use dc:title at all but to take these from a library vocabulary (e.g.
rdvocab). However, my preference is to drop the statement of
responsibility (which doesn't make sense displayed anywhere but as a
continuation of title, which only library software knows to do) and put
title and subtitle into dc:title.
DC still doesn't have a clear "date of publication" but other
vocabularies do. I have no idea if there is one in bibJSON.
On 2/21/13 11:51 PM, Peter Murray-Rust wrote:
> On Fri, Feb 22, 2013 at 4:52 AM, Tom Morris <tfmorris at gmail.com
> <mailto:tfmorris at gmail.com>> wrote:
> dcterms has publisher and date, but unfortunately lacks date of
> publication. It may be possible to find this in BIBO or some other
> bibliographic schema. However, I have seen others place an entire
> publication statement in dcterms:publisher for this exact reason.
> It's a real lack in DC.
> I'd argue abusing the semantics of fields/properties by using them
> to carry information other than what is intended is at best
> prolonging the agony and at worst inviting chaos. If it's not a
> dc:publisher, put the information in myveryown:publication_statement.
> It is impossible to do everything both unambiguously and simply no
> matter what is used.
> DC started with 15 concepts, that could fit on a T-shirt. I've got one.
> There is a "date" field. The date is not precisely defined - TShirts
> have a limited amount of readable area. The idea was to provide
> something that fits most cases.
> BibJSON allows a mixture of namespace and therefore vocabularies.If you
> want to use a term from Schema.org it's possible. It leads for fluid
> documents that are easy to author and may require some processing
> elsewhere if they are to be combined. And this processing may be
> possible and, on occasions may not be unambiguous.
> Again, DC doesn't provide a place for the "statement of responsibility"
> that is in much library data. So the choice is to lose that, or to stuff
> it in title. It also doesn't have a separate subtitle element, so that
> too has to be stuff in title.
> Or define new BibJSON fields to carry them. That's a title, an
> author's name, a series number (?), a translator's name, and a
> parenthetical expression all mushed together in a field that claims
> to be carrying the title. How would anything except a human ever
> make sense of that?
> If there are specialist vocabularies that do this they can be used.
> BibJSON is intended to be easy to author and easy to parse.
> I like JSON as much as the next programmer, but I'm not seeing how
> promoting semantic nonsense benefits anyone. We might as well just
> use BibTex or MARC or one of the other standards that has a long
> history of ambiguity.
> If BIbTEX has a vocabulary that no-one else has, use it in BibJSON. Same
> for MARC.Same for schema.org <http://schema.org>.
> The normal alternative is not to encode the information at all.
> Peter Murray-Rust
> Reader in Molecular Informatics
> Unilever Centre, Dep. Of Chemistry
> University of Cambridge
> CB2 1EW, UK
kcoyle at kcoyle.net http://kcoyle.net
More information about the openbiblio-dev