[libdb-develop] Re: [Bibliophile-development] citation ids?

[Bruce D'A. <bd...@fa...>]

On Nov 29, 2004, at 9:06 AM, Morbus Iff wrote:

> I've mentioned this before, but I don't think /any/ attempt to create 
> an ID that has embedded meaning will work, because any attempt to do 
> that in the past has failed or clashed (ISBNs, UPCs, etc.). I also 
> don't want IDs that are more than 60 characters long, as they're 
> inevitably loaded in a URL, and that makes the URL difficult to pass 
> around.

This is a good point, particularly considering that my current 
stylesheets grab a complete bibliography collection with a single (very 
long!) url!

Still, I don't know how a long string of numbers generated by 
individual dbs helps though.

Documents have citations coded like:

	<citation><biblioref linkend="some-citation-id"/></citation>

This is how they get coded in DocBook, it's how they will get coded in 
OpenOffice, and if I have my way, Word as well (you can embed the same 
citation schema that is going in OOo in WordML).

So how do we make this work in an internet-enabled collaborative 
context?

I got a link to this article about how CiteSeer handles this stuff:

	http://portal.acm.org/citation.cfm?id=1031171.1031275

Someone else posted a comment on the blog about a system that yields 
IDs like:

Gilchrist:NAMS-36-9-1199
Doe:EW-99

The acronyms are for titles: in the first case for an academic journal, 
and in the second a book.  The acronym is constructed by taking up to 
the first four words (other than "the", "a", "of" and so forth) of the 
title.  I like this approach myself.

In Endnote, BTW, there are two fields: one is the database ID, and the 
other is called "label."  People who prefer the citekey approach 
typically use that for their citations instead of db ID.

Bruce