[uf-discuss] Citation Microformat: LazyWeb for BibTeXperts

[Michael McCracken]

Replying here because the thread wandered a bit and I don't have
anything to say about dtaccessed.

On 10/5/06, Brian Suda <brian.suda at gmail.com> wrote:
> Calling all BibTeXperts.
>
> I have found a few free cycles here and there and have pieced together
> the first of many XSLTs that will convert the Citation Microformat to
> various other formats.
>
> I have updated my Straw proposal slightly to avoid collisions with
> class values and to bring it in line with other formats (e.g. hResume,
> what was 'title' is now 'fn') i also added reference to using rel-Tag
> as keywords.
> http://microformats.org/wiki/citation-brainstorming#Example

Great, how about I merge my previous straw proposal as examples using
the same format?

> Taking the implied schema, i began to create an XSLT that maps those
> values to BibTeX. NOTE: this is NOT a 1:1 mapping of BibTeX, it is a
> mapping of COMMON values in the wild to their BibTeX equivalents (or
> atleast i think they are equivalent - some one will tell me otherwise
> i'm sure). Eventually, there will be XSLTs to map between the
> microformat and other citation formats through the Implied Schema.
>
> XHTML-2-Citations
> http://suda.co.uk/projects/microformats/X2C/
>
> So, now we can begin to "round-trip" data. I have marked-up a page or
> two of my own and can convert that to a .bib file. The service is very
> fragile, it isn't nearly as robust as the X2V, but this is still early
> days.

Fantastic!

> I am calling on LazyWeb to check the BibTeX output (i think there are
> some know errors already). As well, i am asking a few people to take
> the straw proposal values and mark-up some examples themselves and
> test the output from X2C as input into BibTeX applications. In
> BibTeX... are there REQUIRED properties? enumerated TYPES?

There is a property list file that BibDesk uses to describe the
'standard' types in bibtex and how they map to other formats here:
http://svn.sourceforge.net/viewvc/bibdesk/trunk/bibdesk/TypeInfo.plist?revision=7561&view=markup

This file is honed from user feedback - we started with what info we
could find from standards and iterated with lots of help from the
users list.

The "FieldsForTypes" dictionary in that file describes the "standard"
bibtex arrangement as documented in the Kopka & Daly LaTeX book (I
think it's that one).

Our terminology goes like this:

@type{key, field="value"}

so when we say the "book" type has required fields of publisher,
title, and year, it looks like this:

@book{key, publisher="", title="", year = 2003}

>The
> Microformats only maps to about 50% of all the BibTeX properties, but
> that should be over 80% of what people are actually publishing, so
> there SHOULDN'T be any issues, but i can't tell unless people test and
> let me know.

I'll look into some test cases and give feedback as I can.

> Is the Straw proposal adequate? Is KEY required (can this
> be accomplied with the ID attribute)?

Yes, as mentioned later on in the thread, the key is required. The ID
attribute is my favorite option for giving a key, although if there's
a UID that might be better.

As a side note, Endnote doesn't produce keys with its bibtex export
and that's raised the ire of many of our users who would like to
round-trip with Endnote users but can't.

> There are two properties that the XSLT does not yet support that were
> in the Implied schema. IDENTIFIERS (which is still under debate) and
> Language, which i COULD pull from xml:lang="en" attribute, but does
> bibtex want "English" or "EN", or should there be a class="language"
> property? or both?

I'm not sure what you mean about language. there's no standard way to
encode in bibtex what language the referenced item is in. If I thought
it was important to include, I'd put it in the "Note" field, which is
usually printed out by style files.

I wrote a long email recently about details of bibtex's non-standard.
I think I'll post it on my site today and point there from here. The
exec. summ. is that the data model for bibtex is an entirely de facto
standard defined by the commonly used style files, and there's only
really a standard for syntax. (But even there it's not formal).

-mike

-- 
Michael McCracken
UCSD CSE PhD Candidate
research: http://www.cse.ucsd.edu/~mmccrack/
misc: http://michael-mccracken.net/wp/