[Bibjson-dev] BibJSON Goals

Thu Jun 2 16:24:07 UTC 2011

Mark MacGillivray <mark at cottagelabs.com> wrote:

> Couple of things to point out:
>
> when converting from BibJSON -> BibTex -> BibJSON as Peter mentioned,
> you would not actually get vaild BibTex out - e.g. there would be
> things in there that are not BibTex. Because we are embedding metadata
> in the first object of the list, and that metadata will include things
> we introduce in BibJSON such as "namespaces", it would not strictly be
> BibTex. 

There is no problem puttine metadata into BibTeX. You just stuff it in a comment
field. There is metadata in the top entry of my bibserver.bib 

http://www.stat.berkeley.edu/~pitman/bibsever.bib

which generates the title and other information displayed in

http://bibserver.berkeley.edu/cgi-bin/bibs7?source=http://www.stat.berkeley.edu/users/pitman/bibserver.bib

The role of BibJSON in this context is to provide an intermediate format  between the BibTeX and the BibServer
so that BibServer can run over any modest sized and well structured biblio dataset.

> We could drop this metadata in a conversion to BibTex, 
> but of course then we could not convert back into BibJSON without losing content. 

NO!  Just put the metadata it in a comment field, or otherwise hide it, without loss.
See e.g.  http://www.netlib.org/bibnet/authors/b/
and similar directories for a large number of personal biblios in BibTeX with high quality metadata
hidden in a BibTeX entry behind tex comments %%%.
There is all kinds of bad practice like this. We need to set a good practice with BibJSON, and provide a
few simple ways like what I did and what Nelson Beebe has done to embed metadata in BibTeX files.
It should BTW be a good and simple exercise for us to convert all of Beebe's netlib to BibJSON, and to provide
BibServer functionality over each bib in his collection, as well as a BibServer index of the whole collection
obained by pulling all the metadata files and making them into a new collection.  This is the kind of use case
we must accomodate with BibJSON/BibServer.

> Not sure what the web service part was actually referring to but yes
> as Sam says using jQuery allows us to build some nice tools for people
> to pass metadata into our bibsoup. This is why I am interested in
> building the BibServer front end in javascript, even if it secondary to python running a front end.

Yes, definitely interesting. About this and BibServer front end in javascript, should keep in contact with the SIMILE 
project and successors see e.g.  http://citeline.mit.edu/
Its hard to say why this project hasnt taken off, but any javascript front end to BibJSON should be able to build on this.
Beware though that I had a developer try to do this a couple of years ago over CouchDB and they failed miserably. Once bitten, twice shy.

--Jim