[tcs-lc] policy for relationships in TCS

Sun May 15 10:39:43 PDT 2005

On Fri, 13 May 2005, Richard Pyle wrote:

> >    The TCS allows more choices if an expert simultaneously authors
> > concepts AND asserts concept relationships. In that case, he or she has
> > the option of placing the relationships inside the newly authored
> > TaxonConcepts OR outside in the RelationshipAssertions.
> >
> >    We recommend the former solution (1) if the author wishes to make the
> > relationship to another concept an integral part of the new concept
> > definiton.
>
> I thought the distinction of where to put Relationship information
> (TaxonConcept vs. RelationshipAssertions) was about whether the
> relationships are "definitive" (i.e., serve as part of the definition of the
> authored concept), or whether they are intepretive (intepreted by a third
> party).
>
> To me, it seems like it should be stronger than a "recommendation" that such
> "definitive" relationships be included within the TaxonConcept part.

Yes, but the real issue here is that workers following the tcs approach
crafting new treatments should be careful to differentiate unambiguously
between definitive relationships and interpretative relationships.

It is common in a taxonomic treatment to indicate the relationship between
the taxa being treated and taxa recognized in earlier treatments.
Typically these are the author's interpretations of how his or her new
concepts relate to concepts of earlier workers, but are not definitive in
that the new author could well come to understand the older work better at
some future time without this having one bit of impact on his or her
concept.  Indeed, I think it will be relatively uncommon for an author to
want his or her concepts to necessary depend on other concepts in any way,
other than in sharing the type specimen.

I believe the default when interpreting the published work of an author
should be to place all relationships outside the concepts. In tcs
discussion we occassional reference the Koperski et al. 2000 German moss
monograph because of its explicit use of concept relationships. Therein
Koperski et al. indicate the relationships of their concepts to concepts
of prior workers.  It is my interpretation that when we markup the
Koperski et al. work in tcs, all of their concept relationships should
fall outside the concept definitions as these are simply interpretations
of past workers and do not define the concepts accepted by Koperski et al.

> > Note that we are not necessarily talking about parent-child
> > relationships.
>
> I understand and agree.  However, I would still find it helpful if someone
> could explain to me what a "parent-child" relationship is in the context of
> concepts.  In the context of NameObjects, I think it has a place in
> establishing hierarchy (although perhaps best using different terminology).
> But in the concept world, I don't understand the distinction between a
> parent-child relationship and an "includes/included in" relationship.  From
> a concept circumscription perspective, it is immaterial whether the
> expressed relationship of "Name1 Sec. Smith > Name2 Sec. Jones" involves a
> case where "Name1" and "Name2" are at the same nomenclatural rank, or
> different nomenclatural ranks.  This is partly why I think that "Rank"
> doesn't really belong within TaxonConcept at all, but should instead be
> inherited from the corresponding NameObject.

I tend to agree, but this is a separate discussion. (I personally think
that rank should be an attribute of a nameObject only and not of a
conceptObject.)

> > Earlier, external concepts may be used as well to nail
> > down the meaning of a newly published concept. Example: Ranunculus
> > abortivus L. sec. Kartesz (2004)  = =  Ranunculus abortivus L. sec. FNA
> > (1997), where Kartesz (2004) places this relationship into the
> > TaxonConcept definition to indicate that the newly published 2004
> > concept is defined (!) by its relationship to the FNA 1997 concept.
>
> Congruent, right?  If so, then why does Kartesz's use of the concept rise to
> the level of a "defined" concept, rather than simply a usage?

So you now see more clearly that there are rather few cases where we want
to place a relationship inside a concept definition.  Kartesz could define
his concept equal to FNA, but then why not just use the FNA concept as
this would be simpler.

> >    We recommend the latter solution (2) if the author does not
> > think of the relationship as an integral part of his or her new
> > concept definition.
>
> I think I understand your point here -- but I wonder, how often would if be
> obvious to the external data-capturer whether a new conept's author intended
> an expressed relationship to another concept as representing part of the new
> concept definition, as opposed to an incidental extra piece of information?

As I explained above, we need to take the conservative approach of
assuming all stated relationships are simply incidental interpretations
unless there is an explicit statement to the contrary.

> > Perhaps other data (diagnosis, included concepts, etc.) are sufficient
> > to specify the present meaning. Or the relationship to earlier concepts
> > is not so clear as to nail down the new definition exactly how the
> > author wants it to be. Example: Equisetum hyemale L. ssp. affine
> > (Engelmann) Calder & R.L. Taylor sec. Weakley (2005)  >  E. hyemale var.
> > affine (Engelmann) A.A. Eaton sec. Radford, Ahles & Bell (1968), as
> > asserted by Weakley (2005). Here the latter author does not wish to
> > define his new concept via its (inclusive) relationship to the 1968
> > concept. Rather, Weakley intends to provide readers with a "guide" to
> > understanding the taxonomic legacy. The precise definition of his new
> > concept lies in the diagnosis and explanatory comments.
>
> Hmmm...I'm not sure I follow this logic.  We presumably have these concept
> definitions:
>
> 1) Equisetum hyemale L. ssp. affine (Engelmann) Calder & R.L. Taylor sec.
> Weakley (2005)
>
> 2) E. hyemale var. affine (Engelmann) A.A. Eaton sec. Radford, Ahles & Bell
> (1968)
>
> If Weakly expressly states that his concept includes that of Radford et
> al.'s 1968 concept, why does that not constitute part of the definition of
> Weakly's concept?  Do we assume that Weakley says somewhere that he provides
> this information only as matter of interest, and expressly not as part of
> his concept definition?  Or are there other ways by which we can make the
> distinction?

Let's consider what Weakley is doing in his flora
http://herbarium.unc.edu/WeakleysFlora.pdf

For each taxon (concept) he recognizes, he wants to relate it to all the
other major works over the past century, in much the style of Kopersky et
al.  This is very helpful to the reader for many reasons you can imagine.
His knowledge of the inner workings of the minds of the many systematists
over the past century who have treated some of the ca 4500 taxa included
in his work far exceeds mine as his reader, but his understanding is still
imperfect.  After publication he may well come to have a different
understanding of what J.K. Small 1933 intended with respect to Aus bus,
but most likely this will not change at all Weakley's concept of Aus bus
and thus should not require him to publish a new concept.  Similarly, if
Weakley were to drop dead today and I was reviewing his work and
discovered an old manuscript of JK Small Weakely had never read, I might
come to understand that a relationship Weakley suggested is invalid
without
it having any relationship to the altogether valid concept he published.

> I understand that Weakley may not correctly understand the concept
> circumscription of Radford et al., and therfore the character-based
> definition may be in conflict with his assertion of the relationship between
> his concept and that of Radford et al.  But that's a problem inherent to the
> entire schema (i.e., would apply regardless of whether Weakley intended the
> stated concept relationship to be definitive for his own concept, or not).
> The same can be said when both character data and voucher specimens are
> provided as part of a concept definition (i.e., some of the cited vouchers
> may not be consistent with the cired character data).

Correct.  But, until July we have a choice as to whether relationships
embedded inside the concept are definitive of the concept or simply
appeared at the same time and place.  What Jessie, Nico, Roger, Robert and
I are proposing is that relationships inside the conceptObject are
definitive, and consequently most relationships should be placed outside
the conceptObject.  This is not altogether intuitive to the reader of the
tcs, which is why we distributed this interpretation.

>
> >    No matter what option an author chooses at the time of authoring a
> > concept, the possibility of authoring another (now by default)
> > third-party relationship at a later time remains. Such a reassessment
> > would "coexist" with the earlier relationship.
>
> I think I follow your point here, but can you be more explicit about what
> you mean by "at a later time"?

Weakley publishes concept Aus bus sec Weakley 2005 wherein he states
congruence with Small 1933 among others.  In 2006 he changes his mind
about congruence with Small 1933 and publishes a new concept.

We might also ask the consequence of Weakley 2006a publishing a
relationship that is contradictory to a relationship embedded in the 2005
conceptObject and whether this publication of a relationship might to some
observers seem as a cryptic publication of a new concept.

> >    We think that this convention of assigning concept relationships
> > inside vs. outside of the TaxonConcepts section is sensible. It is
> > something that experts will appreciate as guidance when defining
> > concepts and their relationships.
>
> I don't mean to be argumentative, and I could easily be brought into
> agreement with you -- but I'm still not sure I fully understand all of the
> implications.  Apologies for the long note, and further apologies for the
> fact that I am leaving for two weeks, and may not be able to follow up on
> the discussion.

Hope this helps, and that by the time you return in two weeks my mailbox
will be empty and the issues will be resolved.

Cheers,
Bob

 ====================================================================
 Robert K. Peet, Professor              Phone:  919-962-6942
 Department of Biology, CB#3280         Fax:    919-962-6930
 University of North Carolina           Cell:   919-368-4971
 Chapel Hill, NC  27599-3280  USA       Email:  peet at unc.edu

             http://www.bio.unc.edu/faculty/peet/
 ====================================================================