[SEEK-Taxon] Taxon Exchange Schema and GUIDs

Kennedy, Jessie J.Kennedy at napier.ac.uk
Tue Jun 15 07:14:27 PDT 2004


Hi Folks
 
You'll probably have seen my reply to Aimee by now...so just to say I'm not yet convinced we should have GUIDs for edges as a general approach to the problem.
 
The basic reason is I would expect things that have GUIDs to be meaingful real world entities that we wish to refer to. I'm not sure that we want to identify every relationship between concepts as things of importance in their own right. 
 
So sorry for the lack of discussion on the GUID issue earlier but I guess we have to discuss this further in the context of what constitutes a concept.
 
As you'll eventually see from the wiki page there is still discussion about whether or not we need to have names as distinct entities form concpepts - but we're working on trying to show how the propsed schema would work....
 
TDWG have set a deadline of 9th August for distributing proposed schemas to the group for comment and we're trying to meet that deadline.
 
Have been and talked to IPNI now too - only group we haven't been able to have discussion with that we agreed to is APNI.
 
more soon....
 
Jessie
 
 

-----Original Message-----
From: seek-taxon-admin at ecoinformatics.org [mailto:seek-taxon-admin at ecoinformatics.org]On Behalf Of Nico M. Franz
Sent: 14 June 2004 19:25
To: Stewart, Aimee Marian; Seek-Taxon (E-mail)
Subject: Re: [SEEK-Taxon] Taxon Exchange Schema and GUIDs


Hi Aimee et al.:

   awesome job!!! Since I revised the Use Cases and already added some stuff about authoring relationships (your "edges"), I'd endorse your solution to assign GUIDs to them. That said, some of the problems and examples you present so well weren't fully clear to me when I worked on the Use Cases. It's a bit better now, but not crystal-clear either.

   My basic motivation for treating relationships that way was Bob's little drawing at the KU meeting. Berendsohn seemed in favor of this too:

Concept 1
Concept 2
Relationship-Type
Author/Reference of that Relationship-Type
(GUID)

   As you point out, that picture soon gets more tricky, as Concepts 1 & 2 (atomic taxa, unfortunately often turning into large molecules...) should each have their own GUIDs already. Then there's the issue of how far up/down the hierarchy a change trickles (scenario 1 - vertical relationships), and whether one can "make use" of someone else's concepts/relations without creating new, very similar yet potentially unconnected ones (scenario 3 - lateral relationships, possibly with vertical implications...).

   Let's just say: issue 1 has to do with implementing stopping rules, and issue 3 has to do establishing responsible practices of citing vs. authoring. "No Hierarchies can share Concepts anywhere above a difference in the tree." is a very perceptive statement. But it doesn't mean that ITIS couldn't e.g. accept a whole set of concepts from an expert revision "en bloc", for a certain time. Once ITIS "adds value", then we're running into the problems you point out.

   As for the stopping rules, I currently plead the 5th, and promise to work on it some more, with examples like those you present. A bit of that was in my property/constituents distinction in our pre-eScience meetings. Properties are rather immune to added child concepts, constituent-based concept definitions are more susceptible. Different scientists may prefer different solutions depending on the context.

   My current proposed "solution" (really an attitude) is to build as much as possible a system that'll let scientists make the choices of what could/should get a GUID in this concept/relation muddle. There will be some who are quite happy citing someone else's more elaborate concepts without implying anything new/different about their content. Think of it as a mosaic, most people will want express their new views in some parts of the tree but leave others deliberately alone. In VegBank, this is handled through status assignments, which may be a third element that could receive GUIDs, since they have authors, too.

   I think in general things are easier with the print legacy, and possibly harder with the digital merging/versioning of multiple taxonomically overlapping databases. We need to discuss this more and understand it better. GUIDS, well-understood relationships, and responsible citing/authoring practices for concepts, relationships, and statuses will all have to come together as a triple-treat to make it work.

Cheers,

Nico

At 11:05 AM 6/14/2004 -0500, Stewart, Aimee Marian wrote:


Seek-Taxon,

We are having a fundamental problem with the Taxonomic Exchange Schema and the assumption that this is the definition of a Concept, and therefore the element that a GUID is attached to.  

The attached HTML document describes our problems and proposes a solution.  

 <<TES-GUID.html>> 
Aimee

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mercury.nceas.ucsb.edu/ecoinformatics/pipermail/seek-taxon/attachments/20040615/c79f4745/attachment.htm


More information about the Seek-taxon mailing list