[fitsbits] 'Dataset Identifications' postings (digest)

Tue Mar 23 10:04:20 EST 2004

It may be good to clarify the context and scope of what Tom is
proposing (at least my take on it; I won't claim to speak for Tom).

The proposal is to introduce the DS_IDENT keyword as a convention for
dataset identifiers and to define one particular set of values for
this keyword - the ones under the autority of the ADS, i.e.,
identifier values starting with "ADS/".  Anybody who wants to
participate in the use of this convention is free to do so, but will
have to comply with the the rules of that convention, which are:

1. the identifier is of the form "ADS/<observatory>#<dataset>"
2. <observatory> must be taken from the list maintained by the ADS
3. <dataset> values are controlled by the data center or observatory
that bears responsibility for the <observatory> archive
4. that controlling authority, and its successors and assigns, must
guarantee access to <dataset> in perpetuity
5. the keeper of the <observatory> data will provide a specific set of
services that allow identifier verification, harvesting, and access to
the datasets

If someone else wants to define another class of identifiers (i.e.,
other than the "ADS/" class), that is fine, but it would probably be
sensible to make sure that the values and useage comply with IVOA
standards (as the ADS ones do) in order to maximize usefulness and
recognition.

I can tell what they, most likely, will look like for Chandra.
There will be (at least) three groups:
ADS/Sa.CXO#obs/<ObsId>
	Points to a particular observation
ADS/Sa.CXO#defset/<name>
	Points to a specifically defined set of observations
ADS/Sa.CXO#bibcode/<bibcode>
	Points to all information we have for a particular paper

Of course, this begs two questions:
- Can two files have the same DS_IDENT value?
The answer should be yes, since a dataset may consist of more than one
file.
- Can one file belong to more than one dataset?
The answer is again yes.  This may mean that we should allow for
DS_IDn keywords.

(I said "files"; you may read "extensions", if you like)

The question has come up in which headers the keyword should appear.
I would recommend putting it in any and all headers where it is
appropriate - primary and secondary.

Hope this helps,

  - Arnold

Don Wells wrote:
> ...
> 
> From: Thomas McGlynn <tam at lheapop.gsfc.nasa.gov>
> Subject: [fitsbits] Dataset identifications.
> Newsgroups: sci.astro.fits
> Date: Wed, 10 Mar 2004 14:20:18 -0500
> Organization: NASA Goddard Space Flight Center
> Reply-To: Thomas.A.McGlynn at nasa.gov
> 
> There is an effort underway at several of the NASA archives
> to provide a standard dataset identifier for data that
> can be retrieved from the archives.  The initial motivation
> is that when authors publish a paper they will be able
> to specify the data that was used in analysis and systems
> like the ADS will be able to provide links to these data
> in a systematic way from the papers (and vice versa for
> the archives). Currently this is done for a few datasets
> but it's a very manual and labor intensive process.  Although
> the initial impetus is coming from some of the NASA sites,
> we've been talking with the VO efforts and hope that the
> ID will be of general utility.  I've no doubt that if ID's
> become established they will be used in many
> different ways.
> 
> There are discussions still ongoing as to the exact format
> to be used.  It is intended that the overall format will be
> compatible with the identification standards that are being
> discussed in the Virtual Observatory world.  An example ID
> might be ADS/Sa.ROSAT#X/rh701576n00 where the ADS indicates
> the the ADS will provide the high level resolution service,
> the 'Sa.ROSAT' is an observatory identifier, and the
> element that follows the # is observatory specific, but
> should be familiar enough for those who have used ROSAT
> data.
> 
> The question for this group is not so much a discussion of the format
> of the ID.  Rather it was pointed out that if these IDs are successful
> it would be useful to be able to have a standard
> FITS keyword that would indicate the dataset id that the current
> file belongs to.  The keyword 'DS_IDENT' has been suggested.
> Does anyone have objections to this or do they know of systems
> that already use this keyword?  Googling DS_IDENT returns an album
> of Donna Summer's but no FITS references.
> 
> Also, are there any issues the we need to resolve regarding
> the usage of the keyword?  One that comes to mind is whether use of this
> keyword should be recommended only for the primary header of a FITS
> file.  If not then a file may not be associated with a unique dataset
> id.
> 
> I'd appreciate any comments, questions or thoughts on the subject.
> 
> 	Thanks,
> 	Tom McGlynn
> 	HEASARC
> ...
> 
> -- 
>   Donald C. Wells             Scientist               dwells at nrao.edu
>                     http://www.cv.nrao.edu/~dwells
>   National Radio Astronomy Observatory                +1-434-296-0277
>   520 Edgemont Road,   Charlottesville, Virginia       22903-2475 USA
> _______________________________________________
> fitsbits mailing list
> fitsbits at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/fitsbits
> 
--------------------------------------------------------------------------
Arnold H. Rots                                Chandra X-ray Science Center
Smithsonian Astrophysical Observatory                tel:  +1 617 496 7701
60 Garden Street, MS 67                              fax:  +1 617 495 7356
Cambridge, MA 02138                             arots at head.cfa.harvard.edu
USA                                     http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------