[Nhcoll-l] how should we encode vague dates in specimen database?

[Paul J. Morris]

There's some substantive discussion of issues around date
representation in the set of tests for DarwinCore Event terms being
developed by the TDWG Biodiversity Data Quality Interest Group's Task
Group on Tests and Assertions.  These focus on Darwin Core
representations for exchange rather than internal database
representations.

https://github.com/tdwg/bdq/issues?q=label%3ATIME+is%3Aopen

One thing that has become clear in the discussions is that we don't
have a way in the community to distinguish between a low precision date
where the collecting event or observation occurred at sometime within
an interval, and a date range (usually for observations) where the
observation occurred over the entire duration of the interval. 

There are approaches for capturing dates of arbitrary precision, such
as the "--" that Paul mentions, or the "**" that muse often used, or as
Carla mentioned, representations of low precision dates as ranges from
a start date to an end date (as found in both Arctos and Specify).
The ISO date format also allows for representations of dates as ranges
(in several forms including YYYY, YYYY-MM, YYYY-MM/YYYY-MM, 
YYYY-MM-DD/YYYY-MM-DD).   

"Spring 1984" may be represented as a verbatim value and an
interpretation to a range, where the set of months comprising Spring
depend on which hemisphere is involved, and may be able to be narrowed
from information about the collector's travels - any such inference of
should be documented. 

-Paul

On Tue, 18 Sep 2018 01:21:22 +0000
"Schultz, Eric" <eric.schultz at uconn.edu> wrote:

> We are cleaning up entries in a legacy database so that our data can
> be imported into a properly configured collections database.  Exact
> dates will be imported in a text string mm/dd/yyyy.  But how to
> render a vague date referring only to season, such as 'Spring 1984',
> or if only the year was recorded? Is it better to record as missing
> data, record as an erroneous but exact date such as 01/01/1984, or
> place at the midpoint of the season?  Whatever we do we will have a
> 'verbatim' field that preserves the original entry, but we want to
> have a version of collection date that is useful and searchable.
> Thanks for any assistance!
> 
> Eric T. Schultz
> Professor, Department of Ecology and Evolutionary Biology
> Coordinator, Joint BS-MS Degree in Biodiversity and Conservation
> Biology Director, Vertebrate Biodiversity Collections
> Chair, General Education Oversight Committee
> University of Connecticut
> Storrs, CT 06269-3043
> eric.schultz at uconn.edu<mailto:eric.schultz at uconn.edu>
> 860.486.4692
> 
> 


-- 
Paul J. Morris
Biodiversity Informatics Manager
Museum of Comparative Zoölogy, Harvard University
mole at morris.net  AA3SD  PGP public key available


-- 
Paul J. Morris
Biodiversity Informatics Manager
Museum of Comparative Zoölogy, Harvard University
mole at morris.net  AA3SD  PGP public key available