I've discussed this a bit with Rebecca off-list, but thought I may as
well post a note for public consumption.

I'm a social scientist who has gotten frustrated enough with my writing
tools to have been working towards something of a revolution in that
realm.  In particular, I'd like to see XML formats replace the
currently proprietary and binary files of products like Endnote.

A group of like-minded non-experts has been working on a DTD to use
with RefDB ( and other projects (for
example the ConTeXt TeX macro package), but midway through I found out
about MODS and so subscribed to the list.  I've been quite intrigued by
MODS, but I have an observation, and then a question.

The observation is that MODS is currently unsuitable for a reference
manager application (whose job is not just to store references, but to
precisely format them) because it cannot properly store the most
significant source an academic is likely to cite: journal articles.  In
particular, we really need elements for things like volume and issue
number.  There may be other similar limitations, but this is the most

The question is really about the broader suitability of MODS -- or a
MODS-like structure -- to this use.  If we want to be concrete (and why
not?), what should we do with BibX?  Here's the DTD:

And here's information on the DTD for the formatting files that will be
adapted to BibX:

Currently BibX has a three part structure (work, publication and set)
which corresponds to the analytical, publication and series structures
of bib data.  It also has -- following existing formats like BibTeX and
RIS -- a reference "type" element, and subtype.  Where we kind of
stalled is wondering if the MARC/MODS approach of a more limited range
of types (text, sound, etc.) and use of "genre" to specify content is
more sensible and flexible.

The key point to recognize is that the model needs not only to be able
accurately store bibliographic data, but it needs to subsequently get
accurately formatted.  The formatting engine, in other words, has to
know what to do with journal articles, versus books, versus book
chapters.  This is why, I presume, the more specific "type" one sees in
RIS and BibTeX, but am not totally sure if this is necessary or not.

So, why -- if at all -- adopt a more MODS-like approach for this use,
or should we just give up on BibX altogether and adopt MODS?

Thoughts appreciated,