[Taxacom] Data query

Richard Pyle deepreef at bishopmuseum.org
Tue Jun 25 03:27:57 CDT 2013


Well, I didn't really want to go down that path, but since you brought it
up..

 

First of all, we have a service that already does that:

http://zoobank.org/Search?search_term=Chromis+abyssus

 

In other words, there's your URL for the name.  It takes you to a
disambiguation page, rather than directly to the page.  It would be fairly
easy to 'smarten' this service up to catch cases where there is only one
search result and auto re-direct to that page (ala Google's "I'm Feeling
Lucky" feature).

 

But frankly, I wouldn't want to encourage people to link to a service as
though it were an identifier. Yes, I know that in some circles, all URLs
*are* identifiers; but some URLs are more optimally designed for
identification than others.  And as much as I am a HUGE defender of Linnean
names as human-friendly identifiers (being an ICZN Commissioner and all),
more than three decades of experience working with computer databases to
track organism names has taught me nothing if not that scientific names make
for really, really, really, REALLY bad identifiers for computer systems.
The problem is not that text strings as such cannot be used by computers as
identifiers.  Rather, the problems mostly involve bad human behavior when
confronted with a familiar-looking thing.  The temptation to change that
thing is just too powerful. Good Identifiers just don't want to be changed.

 

I fear that if I launched into a more elaborated explanation than that, I
would be typing this email for another hour.  I'd rather add a half-dozen
identifiers each to 60 different ZooBank records than spend another hour
writing email..

 

For now, I'll just leave it as this:  I currently established WikiSpecies as
an "Identifier Domain" (mostly for your benefit, in honor of the many hours
of work you have personally put into adding content to ZooBank) - which
means that I treat the URLs as "identifiers" rather than services.  However,
in a sense, it's more of a service.  I don't really need to establish a
"hard link" of this sort to WikiSpecies, because it's much easier to
establish a "soft link".  The idea of a "soft link" is that instead of
explicitly recording in the database "ZB:
8BDC0735-FEA4-4298-83FA-D04F67C3FBEC maps to WS:Chromis_abyssus", instead I
can simply take the name-string ("Chromis abyssus"), replace the space with
an underscore, and slap that at the end of
"http://species.wikimedia.org/wiki/".  Before displaying the link to the
user, I can ping that page, and only display the link if I get a real WS
page.  In other words, there's no reason for me to explicitly make that
link, because I can make the link on the fly.  This is the advantage of
having a "soft" link.  But there are other advantages to using opaque
identifiers (e.g., DOIs . hint, hint, Rod) that are proper actionable GUIDs
- and that's where the "hard links" come into play. An example of this, that
must be explicitly recorded in the database as a cross-link (instead of
created on the fly as a soft link), is "ZB:
8BDC0735-FEA4-4298-83FA-D04F67C3FBEC maps to WoRMS:345844".

 

OK, 'nuff said.

 

Aloha,

Rich

 

 

 

From: Stephen Thorpe [mailto:stephen_thorpe at yahoo.co.nz] 
Sent: Monday, June 24, 2013 9:35 PM
To: Richard Pyle; Tony.Rees at csiro.au; mesibov at southcom.com.au
Cc: taxacom at mailman.nhm.ku.edu
Subject: Re: [Taxacom] Data query

 

Actually Rich, you might think seriously about creating ZooBank URLs, in
addition to, and which redirect to the current ones based on LSIDs, but
based instead on the name, as Wikispecies does. This would greatly
facilitate automatic link creation from WS to ZooBank ...






More information about the Taxacom mailing list