[tcs-lc] Names as Objects

[Nico Franz]

Hi all:

   The current exchanges are great in many ways but we must find a 
balanced view. I am currently participating in a meeting of ecologists 
and biodiversity researchers who are developing new semantics 
("ontologies") to interpret data stemming from many sources and places, 
in order to resolve their inherent heterogeneities and integrate them in 
complex analytical workflows. The TCS developments are similarly aimed 
at facilitating such metadata annotation and semantic enrichment (what 
do these two names uses actually mean?). We need to get Code-prescribed 
semantic connections among names right but cannot view them so narrowly 
as to shut the door on a much richer conceptual perspective, which more 
often than not is in use anyway (even if were not yet very close to 
modeling it).

   There is no question (as Greg Whitbread mentioned before) that we 
need to make a better case for the upshots of such semantic enrichment 
with pretty much all communities involved ("nobody will publish this 
concept stuff..." - well, let's see about that). In the meantime, the 
TCS will enable new ways and platforms for labeling and connecting the 
products of taxonomic research. Very similar efforts are ongoing in 
other communities interested in long-term, large-scale biodiversity 
analyses. We're trying to catch up now.

Nico

Kennedy, Jessie wrote:

>Yes but if people are really saying that they want to tag things with names and not nominal concepts which is what I think Gregor is saying then it wouldn't be perfect for him nor for James(and I have a couple of doubts lingering but can't quite express them).
>The problem for me is the discussion is really limited in terms of views of everyone - so trying to see what the general consensus is difficult.
>
>TCS was designed as a compromise to try to take on board all we heard when we went around discussing with the groups as to what they understood to be concepts and what they would do with them etc. So we spoke with many people and tried to find a way of looking at the data in a common format that everyone could map their data to. In Christchurch there was a unanimous agreement at the Names/Concepts subgroup meeting on the Saturday (with quite a lot of people there) that specimens etc. should be tagged with concepts and not names. Therefore we came up with nominal concepts to allow people to deal with legacy data where all you have is a name but we know there is an implied concept that could be any one of the existing concepts described. (Arguably it might not be any of the concepts previously but this I would put down to inherent errors in interpretation of concepts when doing identifications).
>
>Now people seem to be saying that they want to use names whether or not it's the right thing to do and so they don't want to have to use concepts. TCS was never trying to say this is the way you need to do things - that's why the schema has so many optional things in it, so you can represent a concept however you think it should be done, but we didn't allow names as top level objects, because we thought you could represent names as concepts and it would encourage the world to start solving the concept problem. In Christchurch we saw many very nice examples of systems that had been built which were based on names - but what I heard in I think all of these presentations even when they didn't want to acknowledge the need for concepts was that they had got so far but then they were having a problem with what the names meant. 
>
>We also were promoting re-use across TDWG schemas and tried to show this in Christchurch with our use of ANY to indicate that we would expect to pick up the schema defined by one of the other subgroups for this part of the schema and at least we would include a GUID and the primary key fields to the objects being described by the other schemas. So for publication we had an Endnote based suggestion but replaced this with any to say that if the publication group were looking at this we could replace this with their standard for citing a reference. Same for Vouchers, Character Description etc. With Names we took the Name element from ABCD which was a type.
>
>The Name element wasn't seen to be good enough so LC were going to remodel this. I haven't seen a suggestion for a new Name type yet just a potential XSD for LC which may or may not be mappable to TCS - we tried to map some of it.
>
>Now we are in the situation that some people want to pass names around - by this do we mean bare names or do we mean with all the nomenclatural history?
>what do we mean by a name? Some people say that certain names are just typos - regardless of where they were published, some names don't need to have a publication associated with them (so how do we ever find out about these new names or suggested corrections to existing names and isn't this tracked by nomenclaturalists?)
>
>to satisfy everyone to we need a name object that is simply the scientific name (similar to what Walter wanted for ABCD), a nominal object that refers to the name but includes all the nomenclatural relationships and a concept object that includes the name object or maybe the nominal object - not sure what it would really need...
>
>Which is why I kept it together so you only need to view the parts of the TCS that are relevant to your needs. In any DBMS systems you don't pass around the whole object all the time you do projections or define views on the relations so you see what you need to. hmmmm...
>
>......>O.K., maybe it would be helpful to refer back to the
>  
>