[tcs-lc] Minor modifications prior TDWG ratification vote

[Nozomi Ytow]

Rich,

do we share the key words "regression test" or "loopback test"?
Since TCS is a schema for exchange, if you export a record via TCS
and re-import the record via TCS, then you should get the same
contents except extra contents not fall TCS' scope.

> I think my real problem is that I still don't understand exactly what
> information is implied by "incertae sedis", and why it would be important to
> capture in an exchange schema.  I agree with Paul -- it is just a convienent
> way to flag a hierarchy jump that skips a rank that you might otherwise
> expect it to include. It's not real information, as far as I can
> tell.

Supporse exporting a record
("Rus vodka", "Smith 1917", incertae cedis) , i.e. Smith 1917 stated
that Rus vodka is not a cilliate but can't give its placement.  What
does regression test give as return through TCS?  Does this incertae
cedis data fall outside scope of TCS?


> Hmmm... I'll have to think about this some more.  If I interpret you
> correctly, you are saying that there is an important distinction (important
> in the sense that it should be accomodated by TCS/LC schema) between "parent
> not provided" and "parent not known to provider" -- is that correct?

The distionction would be between "higher taxon of the next rank (in
authors' sense) is not specified in the publication" and "higher taxon
data does not exist in data source" or "higher taxon data is not
exported from data source" depending on query. 


RP> Suppose "we" (TDWG, TCS, whatever) define "incertae sedis" as I
RP> did in the previous message -- where one of the "root" rank 
RP> levels has been skipped in a direct parent-child relationship
RP> (e.g., OrderName is parent of GenusName).

JY> I can't suppose it because you assume mandatory ranks.

RP> No, I do not.  I'm quite comfortable indicating that the direct
RP>  parent of a species name is a Kingdom name -- no mandatory
RP> requirements for intermediate rank at all.

I'm confused.  Are you talking about NameObject (I suspect from your
mention on "OrderName", "GenusName", although I do not know what you
meen thes looks-like-TCS-element terms), or TaxonConcept?
Arn't you require Kingdom as manatory implicitly?

>  It is not about "mandatory" ranks -- it is about defining what
> "incertae sedis" means.

I do not understand.  There are records having "incetae sedis = ture"
status.  How can make TCS passes regression test with such records?
Incertae sedis implies something incomplete in higher taxon chain.
Why do we need to define the meaning of the status from outsied of
data source?   Do you mean interprete, rather than define?

> Either there is a stable definition
> (e.g., skips at least one "root" rank), or there is a dynamic
> definition, in which case each data provider may define it
> differently in different contexts.  
> Also, it is either a property of a Name/Concept, or it is a
> property of the *relationship* between two names/concepts.

Doese the sentence prior to "Also" is relevant?
The issue is the latter senetence.  Logically it is a type of
relationship, but making incertae sedis as a relationship type
requires to make Relationship/TaxonConcept minOccur=0 which
weaken TCS.  (Why do I need to repeat it three times, Rich?)


> > Why user should provide NameObject?
> Because, by my understanding, "incertae sedis" is defined 
> in the context of ranks, and ranks are properties of NameObjects.

Disagree completeley.  Japanese taxoomists provides Japanese names
to taxa at higher ranks also.  Rnak is unnecessary a exclusive
propary of NameObject.  Note that TaconConcept has Rank element
optionally.

> But I believe that
> "incertae sedis" is itself a property of a parent/child
> *relationship* between two concepts.

Unnecessary between two concepts, because the next-available
higher taxon may ungiven in the publication.  If ungiven,
then incertae sedis is a type of relationship between the 
taxon concept and nothing.


> If there is a parent/child relationship between two
> concepts, and those two concepts are not each linked to
> a NameObject (with associated ranks), then I don't know
> what incertae sedis would mean.

I lost.  Are you saying that there are cases where incertae sedis 
can be calculated, and if incertae sedis is calculable, it is
unnecessary to export ?

> Maybe it simply means "parent concept not known to data provider",

Not "unknown to data provider" but "stated incertae sedis by the
author".  "Unknow to data provider" can happen when author did
not specify the higher rank, but it is not incertae sedis except
the author stated explicitly (either as incertae sedis or in
natural language).

> in which case
> it would have no connection to ranks, and therefore no necessary involvement
> with NameObjects, and it would be a property of a Concept (as opposed to a
> property of a relationship between two Concepts.

I lost again, but anyway, it is not property of NameObject.
  

> I don't thinkincertae sedis should be, either -- because
> I do not understand what actual information is being conveyed by it.

It conveys that it is stated so by the author.

> Suppose my database has records for 10 species-concepts,
<snip example description>
> As a data provider, should I, or should I not, flag the first 5 records as
> "incertae sedis"?

It's up to you, but I wouldn't flag except if the author stated so
explicitly, and it is unlikely in your example.

> If I do flag them as such, what information have I
> provided in doing so?  If I do not flag them as such, what useful
> information have I failed to provide you?

If the author didn't state as incertae sedis explicitly, you lose
nothig withoug flagging.

Cheers,
JMS