[tcs-lc] You don't need embedded names to do concepts

Richard Pyle deepreef at bishopmuseum.org
Thu Mar 10 13:17:04 PST 2005


Jessie wrote:
> > give them GUIDs before we create concepts that we
> > need GUIDs for - 2 GUID spaces to deal with,

Roger responded:
> This is only a disadvantage from the point of view of
> a concept only world. People who want to refer to Names
> want two GUIDs. It is an advantage from their perspective.

Don't count me among those people.  At the TDWG meeting at Christchurch, the
one key take-home message of my presentation was a slide with the title "One
GUID to Rule Them All" (quoted from the famous trilogy, "Lord of the
GUIDs").  I do *not* advocate separate GUID spaces for names and concepts.
In fact, I swing the opposite direction in my desire to have a single GUID
space for all Name-Usage instances (including Identifications, etc.). See:
http://www.tdwg.org/2004meet/EV/TDWG_2004_Papers_Pyle_1.zip

In Taxonomer, I took the view that the GUID of the Name-usage instance that
represents the Code-compliant original description of a name could serve as
a surrogate GUID handle to the corresponding name object.  Ironically, this
is the view (more or less) that I am now arguing against.  In Taxonomer, I
always had this queasy feeling about a "dual-meaning" GUID -- i.e., that in
some cases the GUID referred to the object represented in human-readable
form as "Aus L.", and in other cases it referred to the object represented
in human-readable form as "Aus L. SEC. L.".  This meant that software-level
business rules were necessary to distinguish the two (actually, in a
relational schema subtyping is straightforward, so I actually could impose
restrictions at the schema level).

However, once I realized that there could be an object (with its own GUID)
that would be represented in human-readable form as "Aus L. SEC. [Nobody]"
(i.e., the TCS Nominal Concept instance) my queasiness disappeared entirely.
To me, this epitomizes "elegance" in solving a database conundrum.

So obvious this is to me, that I was absolutely convinced that the only
reason everyone else didn't see this as obvious had to do with some sort of
communication failure on my part.  As should now be evident, my natural
response to a perceived communication failure is to simply INCREASE THE
VOLUME! ;-)

At this point (with the benefit of a good night's rest), I now see that many
of my ramblings were probably counterproductive.  Couple with my need to
fulfill other obligations (yes, I do have a life outside of email), I now
intend to dramatically tone down the rhetoric.  Don't misinterpret this as a
reduction of my enthusiasm for this topic of conversation -- just a
realization that quantity of email doesn't trump quality.

:-)

But before I sign off on this note, a couple points of clarification:

Jessie wrote:
> > If we agree people sould use concepts even if they are (in my terms)
> > nominal concepts we need to create the equivalent nominal concepts for
> > every name and the original concepts for every name as well as the
> > revised ones.

I agree completely with this statement, and I am convinced that a nominal
concept in my terms is identical to a nominal concept in Jessie's terms
(from a concept-definition perspective).  The only difference is that in my
view, a Nominal concept instance will have embedded within it the
nomenclaturally important non-concept name-object data.  If you strip that
data, you have a TCS Nominal Concept as it currently stands in TCS.

Roger wrote:
> Original concept circumscriptions are of no more importance than any
> other circumscription - possibly generally of less importance - other
> than the fact that they create the name and attach it to a type.

Exactly!  The nomenclatural acts contained in an original description are of
fundamental importance to all subsequent applications of names, but the
concept definition contained in an original description is no more special
than any other concept definition (except that it cannot ever be regarded as
a misidentified or misapplied name).

Jessie wrote:
> > So I think it will put people off ever getting around to
> > concepts for a long.... while....

Roger Responded:
> Here you have a point and it is the one thing that worries
> me and I would like ideas as to how we encourage people to
> use concepts but not mandate them to do it. I think that
> is the challenge.

It worries me too, and my solution is to embed nomenclatural data within
Nominal Concepts, thereby "forcing" all users who wish to attach data to a
name-object directly, must do so by passing through a concept object (i.e.,
the Nominal Concept instance for the name).

> At which point I will stop.

Ditto.

Aloha,
Rich





More information about the Tcs-lc mailing list