On ID "sameness"

[Rob Seaman]

This is an interesting discussion to those of us in the peanut gallery.
Two comments.

First, Arnold's use case implicitly assumes a one-to-one, or at most,
a one-to-many mapping between derived data products and a raw data set.
There may be several variations of a particular observation:

Event list          SAO:CDA:CXO:2000:primary:acisf02000N001_evt.fits
FITS hi_res image   SAO:CDA:CXO:2000:primary:acisf02000N001_cntr_img2.fits
Jpeg hi_res image   SAO:CDA:CXO:2000:primary:acisf02000N001_cntr_img2.jpg

...but there may also be derived data products that result from two or
more observations.  I won't presume to tell you how to handle this - and
perhaps the answer is already evident to folks who are spending more
than a half hour a week on this discussion - but please make sure this
many-to-many case is supported.

Second, ground based archives are on a fast track, at least at NOAO.
The sociology of ground based optical/IR astronomy is such that our
archives will contain many data sets that were only loosely curated.
It may certainly be appropriate to set a level of technical buy-in to
NVO that such PI-contributed data products will not be able to meet -
at least not without significant massaging at the point of archive
ingest.  It may be appropriate, but it may not be wise to set the
level of buy-in too extremely high.  Please keep registry and
nomenclature requirements at a level that an individual researcher
and her grad student can afford the time and money to meet.

For instance:

> it does require a fully hierarchical object space.

Sounds like a great thing.  What are examples of non-hierarchical
or partially hierarchical object spaces - and how do they fall short
of the requirements being discussed?  And why should an individual
PI care to learn about such issues?

Arnold also says:

> What I was saying here is that there need to be two lists of
> prescribed (and unique) IDs:
> - Data center identifiers (such as NASA:HEASARC or HEASARC and SAO:CDA
>   or SAO or CDA)
> - Mission or telescope or observatory identifiers (such as CXO and VLA)

These are really three lists.  A mission or a telescope is a scientific
entity - a cluster of facilities and the experiments possible on those
facilities.  A data center is a library or museum or simply a bookkeeping
entity - whatever analogy floats your boat.  An observatory, however,
is a political entity.  NOAO is an observatory that manages many
telescopes and many data centers.

Note that the left hand side of the name space also suffers (or benefits)
from the many to many mapping discussed above.  The Mayall is a telescope
managed by NOAO.  The WIYN is a telescope managed by four political entities -
NOAO and three Universities.  Gemini consists of two telescopes managed by
seven observatory entities.  Some of this complexity can be smoothed out
at the data center level - but not in all cases.

Rob Seaman
NOAO Science Archive