libdb-develop

[libdb-develop] citation ids?

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

	
I just posted something on my blog about citation/record ids, which is  
an issue that I think is going to become increasingly important.

http://netapps.muohio.edu/blogs/darcusb/darcusb/archives/2004/11/27/ 
citation-ids

Comments?

Bruce

[libdb-develop] Re: [Refdb-users] citation ids?

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

On Nov 27, 2004, at 6:47 PM, Markus Hoenicka wrote:

> Just a random example from Pubmed from our friends devoted to  
> chemistry:
> Studt2004LewisAdductsoftheSide-OnEnd-OnDinitrogen- 
> BridgedComplex[{(NPN)Ta}(2)(mu-H)(2)(mu-eta(1):eta(2)- 
> N(2))]withAlMe(3),GaMe(3),andB(C(6)F(5))(3): 
> Synthesis,Structure,andSpectroscopicProperties
>
> This may look a little contrived but this was actually the first hit
> I've got from a Pubmed query for "nitrogen". There are certainly
> limits to the usability of including the title.

Yeah, I thought of that.  OTOH, you could always just grab the first  
four words of a title.

Blog comments pointed out other issues, though.  I really don't think  
there's an ideal way; that all have compromises.  Still, I wonder if a  
consensus could emerge about best practical approach?

Bruce

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

[Morbus I. <mo...@di...>]

> I just posted something on my blog about citation/record ids, which is  
> an issue that I think is going to become increasingly important.

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.

-- 
Morbus Iff ( you, me, eropuri? aawwwwwWWWw yYeahahhHHAhhh )
Technical: http://www.oreillynet.com/pub/au/779
Culture: http://www.disobey.com/ and http://www.gamegrene.com/
icq: 2927491 / aim: akaMorbus / yahoo: morbus_iff / jabber.org: morbus

[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

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

[Morbus I. <mo...@di...>]

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

Someone may want to poke around in the RDF/SW world for this one. They 
have very similar problems: if I, myself, give me a URI of dc:morbusiff, 
there is nothing stopping another person, unknown to be, calling me 
dc:morbus_iff in their application. How could the RDF/SW ever function 
with these statements that refer to the same thing, only with different 
IDs? The only relevant response I got back was a conversion router: feed 
it one ID, and get back a list of all relevant IDs that are equivalent, 
very similar to the OCLC's xISBN service:

  http://www.oclc.org/research/projects/xisbn/default.htm

-- 
Morbus Iff ( evil is my sour flavor )
Technical: http://www.oreillynet.com/pub/au/779
Culture: http://www.disobey.com/ and http://www.gamegrene.com/
icq: 2927491 / aim: akaMorbus / yahoo: morbus_iff / jabber.org: morbus

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

[<ar...@uw...>]

Bruce writes:

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

and I wonder about candidates for hosting a globally accessible and low 
barrier registry to avoid clashing namespaces. I have been thinking about 
this in terms of trust networks and the use of FOAF (Friend Of A Friend), 
and have been turning to the Open Directory Project (ODP) 
<http://dmoz.org> for  subject URIs (see example at the end). The CiteSeer 
project is pretty amazing, and is very citation-centric, but they have 
been dealing with a lot of system load issues. I also wonder if OCLC might 
also be a possibility, they do something like this for Names already, for 
example: <http://errol.oclc.org/laf/nr2003-18130.html>.

art
---
<rdf:RDF
xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#"
xmlns:foaf="http://xmlns.com/foaf/0.1/"
xmlns:trust="http://www.perceive.net/schemas/20020725/trust#">

<foaf:Person>
<foaf:nick>morbus</foaf:nick>

<trust:trustsHighly>
<foaf:Person rdf:about="#Morbus_Iff">

<rdfs:subject 
rdf:resource="http://dmoz.org/Computers/Programming/Languages/Perl/" />
</foaf:Person>
</trust:trustsHighly>

</foaf:Person>
</rdf:RDF>

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

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

Can I ask that people post further thoughts on the blog?  I'm trying to  
get people from different projects involved in the discussion, and it's  
actually been helpful.

On that, I just had an idea in response to one of those comments  
(thanks Rick!), which is to use existing ids along with the info  
scheme.  E.g. "info:lccn/n78089035".  See:   
http://info-uri.info/registry/docs/misc/faq.html.

So if I'm citing a book, I might do:

	<citation><biblioref linkend="info:lccn/n78089035"/></citation>

.. if a journal article with a DOI:

	<citation><biblioref linkend="info:doi/10.1000.10/12345"/></citation>

I think I like this!  AFAIK, it should be fairly easy to process with  
XSLT; without any significant performance penalty (at least I hope!).

Anyway, blog entry is at:

	http://netapps.muohio.edu/blogs/darcusb/darcusb/archives/2004/11/27/ 
citation-ids

Bruce