[Tcs-lc] Progress on TCS/LC

Richard Pyle deepreef at bishopmuseum.org
Wed Mar 30 22:12:22 PST 2005


> The main issue which is one that we have chopped and changed
> our minds during the development of the TCS is whether or not
> it is more appropriate that names be modelled as objects in
> their own right or be treated as (parts of) concepts as we
> proposed in NZ.

I have to parse the above paragraph to provide feedback:

1. whether or not it is more appropriate that names be modelled as objects
- I believe it is more appropriate to do this.

2. or be treated as (parts of) concepts
- I believe they should (hence, 1 & 2 are not mutually exclusive).

3. or be treated as (parts of) concepts as we proposed in NZ
- Not exactly how you proposed it in NZ, but not terribly different, either
(my view).

> We have heard many arguments for and against both approaches

...as well as arguments from some (e.g., me) championing both approaches
simultaneously.

> and I believe that through the discussions there is now a
> better understanding of some of the requirements of the TCS.
> But it is difficult to know how generally the views are held
> as most people don't have the time that is required to consider
> all the implications of any schema proposed.

Far & away the most useful exercise that helped me understand TCS (and
reshape some of my own ideas -- not all of which have been articulated on
this list yet) was to start creating a real-world instance document.  I
would highly recommend to anyone that they pick their favorite (smallish)
genus and try to represent its constituent species in a TCS document.  I
knew essentially nothing about XML in Christchurch, but I found it easy to
get my head around once I started playing with it a bit.

> It is also difficult to ensure that all issues have been
> considered and it is very tempting at times to be convinced
> that "this one" is the solution but usually someone finds a
> problem. It also seems to be clear that people think that
> there are many ways to model the information and whatever way
> we go I'm sure that someone will feel they cannot express
> what they need to be able to.

I have reasonably high confidence that whatever schema we end up with (even
if unmodified v0.95), we will be able to accomodate the needs of most
people.  The main arguments have been about drifting towards optimality of
information structure.

> So from the discussions we have come to the following conclusions
> Many people believe that names are objects in their own right and
> won't be satisfied unless they can have that.

I hope you don't count me among them.  Although I (still) think it is a
mistake to entangle name-name relationships and
circumscription-circumscription relationships, that doesn't mean I won't be
satisfied with TCS (whatever form it takes).  I'll use it no matter what it
is, because I can see that it will accomodate the information transfer
needs, even as it currently stands.  But I think that further discussion
(and real-world instance document generation) will help us address questions
about optimal structure.

> Names must be seen as different and separate things from concepts.
> Treating names as views (subset of the TaxonConcept element) on
> original concepts does not seem to be an acceptable approach.
> Nominal concepts (concepts with names but no definition) were a
> good mechanism for dealing with legacy data where "identifications"
> were marked up as names where they should be concepts.
> People wanted the schema to be less ambiguous - more precise
> definitions of elements and relationship types. Rules should be
> followed.
> References should be elaborated to include a simple textual
> representation of the target and cater for various types of
> identifiers.
>
> As a result we have decided to put together a version of the
> schema with names as top level objects and we have started work
> on (and continue) this with Roger.

....so why was the idea of treating Nominal-type TaxonConcepts as the
bearers of nomenclatural data rejected?  Some XML technicality? Some
fundamental problem of logic? Some other reason????

> Hopefully this will allow the schema to progress to version 1 for
> voting more quickly. This schema will allow those interested in
> names only to transfer names. Names will only have a short list
> of specified relationships that will be allowed between them,

Actually, I think the list of name-name relationships is quite a bit longer
than the list of concept-concept relationships.

> those explicitly mentioned in LC. Concepts will refer to
> Names but otherwise not really be different.  Anywhere
> that names currently use relationships which are not purely
> nomenclatural will require a concept to be created to allow
> these relationships to be specified.

Ooo....I'm not liking the sounds of this....

I need to see how you make this distinction.

> Roger and Robert will make some final adjustments to this
> version and then Roger will create instance documents to support
> the schema so please be patient and wait until this is available
> for comment. Hopefully Roger's instance documents and documentation
> in creating these will help everyone understand how their data
> would be mapped to the TCS and we can then see if this is acceptable
> to all.

Sounbds good, but.... should I continue my work generating an instance
document for v0.95?  Are you definitely abandoning that version in favor of
the one Roger & Robert are working on, or is that newer version more of a
"what if" scenario?

Aloha,
Rich




More information about the Tcs-lc mailing list