property:combinationExAuthorship

[nielsklazenga]

combinationExAuthorship (property)

Label	Combination Ex Authorship
Definition	People the combination has been attributed to, but who didn’t validly publish it.
Usage notes
Comments
Required	No
Repeatable	No
Constraints	Literal

[nielsklazenga]

@ghwhitbread : Thanks for clarifying that "ascribe" applies to "ex" authorships. So, in your example, wouldn't you also regard the authorship of the protologue to be "Nutt. ex. Seem."? Are there specific rules in the Code that essentially say "the authorship of the name is different from the authorship of the protologue"? The rules in zoology are essentially the same in such "ascribed" cases (if that it what is meant by "ascribed") -- the only difference being that we would reverse the sequence of authors around the "ex" (i.e., it would be "Seem. ex. Nutt."). We don't define "protologue" in the zoological Code, but my understanding of the meaning of that term in the botanical sense is that it is the collective set of required information necessary to establish a name in the sense of the Code. In this context, the authorship of the protologue may be different from the authorship of the publication containing the protologue, but I was unaware that the authorship of the name could be different from the authorship of the protologue.

@nielsklazenga : I didn't see any indication in the link you provided for how to determine the authorship of the protologue (that word appears only three times in the linked page, none of which address how to treat the authorship of the protologue).

I see three potential authorships:

1. Authorship of the publication
2. Authorship of the protologue
3. Authorship of the name

We all agree that 1 and 3 can be different. What I'm not clear on is whether 2 and 3 can be different from each other. If, in botany, they can be different, does that mean there may be potentially three different sets of authors? If so, when the authorship of the protologue differs from both the authorship of the containing publication and the authorship of the name, how is the authorship of the protologue defined? If there is no clear definition, then isn't it logical to assume that the authorship of the name effectively represents the authorship of the protologue (with or without "ex" authors)?

Originally posted by @deepreef in https://github.com/tdwg/tnc/issues/84#issuecomment-754320346

[nielsklazenga]

@deepreef, we do not talk about 'protologue' like that. A protologue is just everything that is associated with a name when it is (validly) published (https://www.iapt-taxon.org/nomen/pages/main/glossary.html#protologue), so the protologue is the publication (or part of the publication that deals with that name).

[deepreef]

Thanks, @nielsklazenga : and yes, that's exactly my point. However one defines "protologue", it (and its authors) are an instance of Reference (publication), not "Name" (or TNU). Thus, I prefer an approach that treats a protologue as a "microreference" (granular publication instance), and assign fully parsed authorship (and date) to that instance, from which the complete/canonical name authorship can be constructed according to a defined template.

In that context, I would prefer to have a single, unparsed "nameAuthorship" term associated with a name instance, and not worry about defining some number of semi-parsed authorship terms.

[nielsklazenga]

Also, 'ascribed' does not necessarily apply to 'ex' (or v.v. rather). In @ghwhitbread 's example, Nuttal is the ascribed author and Seemann is the publishing author, but, if Seemann had indicated anywhere in Flora Vitiensis that Nuttall made a contribution to the descriptions, it would have been Gossypium tomentosum Nutt. rather than Gossypium tomentosum Seem. or Gossypium tomentosum Nutt. in Seem. rather than Gossypium tomentosum Nutt. ex Seem.

[deepreef]

Thanks, @nielsklazenga -- I understand that, but my point was that "authors" correctly apply to text included in a publication, which goes beyond simply the "name". Therefore, in my view any parsing of authorship strings (to any degree) should be part of a discussion of authorships of publications (or "micro-publications", such as protologues); not authorships of "names". We often represent authorships of "names", which is why I think it would be OK to have a single, unparsed "nameAuthorship" term that can be applied directly to a name. But any parsing of authorship strings should be applied to the reference instance associated with the establishment of the name -- not of the name instance itself.

[nielsklazenga]

No argument here. That's what I have been saying right from the start.

[deepreef]

Ok, so we agree that for the names, we only need nameAuthorship (full string), and not basionymAuthorship, combinationAuthorship, combinationExAuthorship, or any other such terms for "semi-parsed" authorship information?

[nielsklazenga]

Yes, we do. We already agreed when we first talked about it over two years ago.

[ghwhitbread]

... there are use cases based on only names with a need for parsed authorShip terms - so also better if they are properties of the name.

[mdoering]

I agree with @ghwhitbread it would be better to have parsed name authorships attached to names directly in some cases, e.g. pure name parsing. As long as the entity types of such parsed Authorhip is the same as for references I don't think that hurts much. It allows us to refer to the same authorship from both a publication and a name.

[deepreef]

@mdoering -- I thought you had previously said that the partial parsing was of limited value (favoring either a single full-text term or a fully parsed set of terms)?

OK, so if @ghwhitbread and @mdoering think there is value in having partially parsed terms (attached to Name instances), then we need to decide what those are to be. Do you prefer the "meaning" approach (originalCombinationAuthorship/basionymAuthorship; subsequentCombinationAuthorship, combinationExAuthorship, etc.), or the "formatting" approach (ala ABDC: authorTeamOriginalAndYear, authorTeamParenthesisAndYear, combinationAuthorTeamAndYear, etc.)?

But can I also confirm that both of you (and anyone else following this discussion) agree that "Authorship" information, especially in a fully parsed form, is properly applied to an instance of Reference (e.g., Protologue), rather than an instance of Name? Do we all acknowledge that the so-called "authorship" of scientific names is best/most correctly inherited from the authorship of the Protologue/Publication instance (i.e., Reference connected to the TNU that represents the nomenclatural act for which authorship is applied) -- and that the only reason we need any authorship terms at all for Name instances is because that's how the vast majority of "dirty" data represent it?

[deepreef]

OK, thinking about this more, @mdoering wrote:

It allows us to refer to the same authorship from both a publication and a name.

I guess my entire point of all these posting is that I don't believe there is such a thing, in an informatics or conceptual sense, as "authorship of a name". Obviously, we think in terms of "authors for names" a lot, because the Codes insinuate it, and there are widely used conventions for placing names of authors next to scientific names. But in an informatics sense, the authorship applied to names is inherited from the authorship of the publication in which the name (or act) was proposed/established.

The ICZN Code defines it this way:

"The author of a name or nomenclatural act is the person who first publishes it [Arts. 8, 11] in a way that satisfies the criteria of availability..." (Art. 50).

Where the confusion comes in is when the authorship as applied to the name doesn't match the authorship of the larger publication that contains the new name or act. The ICZN Code addresses this in Art 50.1.1:

"However, if it is clear from the contents that some person other than an author of the work is alone responsible both for the name or act and for satisfying the criteria of availability other than actual publication, then that other person is the author of the name or act."
[emphasis added]

The bit about "satisfying the criteria of availability" effectively means the "author of the protologue".

Here is the reason I'm making a big deal about this. We currently live in a less-than-ideal world where most of our data are "dirty" (= not fully parsed or normalized). It's fine that we design our data exchange standards around this reality, which is why I'm fine with various authorship terms applied to instances of TaxonomicName -- as a practical/real-world solution to dealing with the large mass of "dirty" data.

But for those of us who operate on the (currently) rarified realm of fully parsed and normalized data, I want to ascertain that "authorship" (the link between agents and the work they produce) applies to instances of Reference, and that TaxonomicName instances inherit authorship information from their corresponding publications. I have never found an exception to this, as long as we acknowledge something akin to "protologue" as a (more granular) instance of Reference -- allowing us to apply the specific authorship of a TaxonomicName as being the authorship of the protologue (including "ex" authorships, "in" authorships, date/year, and all the other important bits).

In other words, do we all agree that any authorship terms applied to a TaxonomicName class are "convenience" terms necessary for dealing with "dirty" data? Or, would a fully parsed/normalized data model require a structure for applying authorship values directly to instances of TaxonomicName, independent of whatever authorship values are applied to associated Reference instances (accommodating something like 'protologue' as a highly granular instance of Reference)?

If we agree, then I think it's important that the wording of the definitions for these authorship terms as applied to TaxonomicName instances make it clear that their values are defined as representing authorships of the associated Reference instances (whether publication as a whole or, in cases where the protologue authorship differs from the containing publication authorship, then authorship of the more-granular protologue Reference instance).

Again, I apologize for harping on this, but I really feel we need to adequately future-proof whatever standard we come up with, and in the case of authorship, I want to find out if we're all on the same page for how the information should be structured in an idealized implementation.

[mdoering]

@deepreef I think that is correct in an ideal world. Question is what TCS2 is supposed to be. If we inherit authorship from a reference,

• namePublishedIn is all we need for structured data, see tdwg/tnc#87
• it fits the Plazi style Treatment model. An authorship for the treatment is the granularity we need. It might then be good to require a Treatment instead of any Reference for namePublishedIn
• it requires TCS to properly define a standard Reference class
• we should have a verbatimAuthorship or sth like that for the entire unparsed string which might contain all kinds of extras

This feels like a major change though and I would prefer to remove all redundant terms if we follow this path. Or at least prefix them all with verbatim as DwC does.

[deepreef]

Thanks, @mdoering ! My feeling for TCS2 is that it should be designed to accommodate the real world, but lay the groundwork for a future TCS3(?) via a graceful transformation to something closer to the ideal world. We do NOT need to force inheritance of authorship from Reference in TCS2. But we should accommodate it (when available), and we should acknowledge in documentation for the "denormalized" terms such as Authorships (and dates?) as applied directly to TaxonomicName instances.

Incidentally, PLAZI treatment = TNU (or, more correctly, there is a clean relationship between a single Treatment instance and a single TNU instance), so I think we should focus on TNUs as the primary unit of data linking/exchange.

Bottom line: no major change needed -- just a subtle acknowledgement/understanding in the associated documentation.

Incidentally, I also thought about something like verbatimAuthorship -- but I think that is overkill. I can explain in detail why I think this, if you like.

[deepreef]

Just for further clarification: stuff associated with "authorship of protologue" is, I think, out of scope for TCS2. We should address it in a more Reference-focused context, which TCS2 can leverage, but not need to define.

[ghwhitbread]

@deepreef. I thought we had earlier dismissed the management of treatment TNU ( and names) as references ( which they could easily be). I prefer not so we can use a more granular definition of TNU to include usage within a treatment. But if we continue in the direction of usage = treatment = taxon concept ... why not?

However the ICN accepts authorship at the level of “description or diagnosis” within a protologue, and even permits ascription by reference to an earlier effectively published diagnosis in another work. Authorship then is the author of the part of a work that is the “name”, even if technically (according to ICN) not part of a name. Authors with names definitely makes for a more tractable reference object.

[deepreef]

@ghwhitbread : Yes, we are in agreement. Just to be clear: a TNU is not a Reference. But a PLAZI "Treatment" (which is a unit of publication) has a 1:1 correspondence with a corresponding TNU. Note the emphasis on "a" (singular). Yes, there may be many TNUs included within a single treatment, but only one of them represents an assertion of an accepted taxon (the other TNUs represent synonyms as asserted by the Treatment, which could be though of in some ways as "child" TNU instances linked directly to the "accepted" TNU that corresponds directly to the Treatment). Incidentally, I don't think of the synonym TNUs as "concepts" -- only the "top level" (accepted) TNU is a candidate to represent the basis of a "taxon concept" (in my view).

ICZN has similar rules to what you describe for ICN. I think it still works OK for authorship associated with a Reference instance (whether or not you call it a "protologue"), and inherited by a TaxonomicName instance (in an ideal world).

[nielsklazenga]

We could do this:

• basionymAuthorship
• newCombinationAuthorship
• basionymAscribedAuthorship
• newCombinationAscribedAuthorship

This resolves the problems that at least I had with the parsed authorships: it being semantically empty and incorrect use of existing terms. newCombinationAuthorship rather than combinationAuthorship, because a 'combination' is a genus name with one or two epithets, so basionyms can also be combinations and new combinations are not necessarily combinations. The latter is not actually true as we would not call a uninomial 'comb. nov.' but 'stat. nov.'. newCombinationOrStatusAuthorship might get a bit too pedantic though.

EDIT: Hang on, it would have to be newTaxonOrNameOrCombinationOrStatusAuthorship. Maybe just combinationAuthorship then and tell people not to parse the term.

[mdoering]

my thinking and how its implemented in GBIF, COL and ChecklistBank is that basionymAuthorship only exists when you have a subsequent combination and as such brackets around the authorship. The authorship of the original name is placed in the combinationAuthorship - the authorship for the exact combination you are looking at. In zoology always empty in case of "new" combinations.

Btw, in ColDP we have also started to allow list of author/agent instances to be listed as the authorship, not just strings.

[nielsklazenga]

@mdoering your comment crossed with the edit I just made to my comment. We are in agreement. I will get over the fact that combinationAuthorship is not quite correct (in my mind at least). The suggestion to use 'Ascribed' rather than 'Ex' stands.

I was going to mention the Author Agents next. IPNI and Tropicos have had them for years. So, we should have it. I am assuming we do not want roles here?

[mdoering]

combinationAuthorship clearly makes no sense for uninomials, but sth will never be right ;)

[nielsklazenga]

Yes, I'll get over it. Just going to put it in. Maybe someone will come up with a better term during the review.

So, just author for the Agent authors? I thought we might just put it in, even though it is not strictly in TCS 1.

[mdoering]

So, just author for the Agent authors? I thought we might just put it in, even though it is not strictly in TCS 1.

yes! Just not sure from the top of my head how an ordered list needs to be expressed...

[nielsklazenga]

Just not sure from the top of my head how an ordered list needs to be expressed...

We'll leave that for application profiles (or applications really).

Vocabularies can be enhanced by building upon this basic term layer. Added features may include:

• constraining the use of literal-value terms through data types
• establishing relationships among class terms by applying a domain model
• introducing additional term properties that generate entailments that can be materialized through machine reasoning

In aggregate, these enhancements may take the form of an application profile or formal ontology. Unlike the basic layer,
which is broadly applicable, vocabulary enhancements may be used by a subset of the community whose interests are served
by the constraints imposed by the enhancements. Thus there may be any number of enhancements developed for a
particular vocabulary.
https://github.com/tdwg/vocab/blob/master/vms/maintenance-specification.md

[nielsklazenga]

There is actually two ways we can do it.

Either:

• basionymAuthorship
• combinationAuthorship
• basionymAscribedAuthorship
• combinationAscribedAuthorship
• author (for the Agent authors)

Or:

• basionymAuthor
• combinationAuthor
• basionymAscribedAuthor
• combinationAscribedAuthor
• basionymAuthorLiteral
• combinationAuthorLiteral
• basionymAscribedAuthorLiteral
• combinationAscribedAuthorLiteral

I think I prefer the second option, which is also closest to what IPNI and Tropicos are already doing.

[nielsklazenga]

Merging with #239.