On ID "sameness"

[Arnold Rots]

I think I'm seeing the light.
The solution may lie in imposing a logical mapping between objects,
name spaces, and IDs, leaving that mapping up to the issuer of the
object.  As far as I am concerned we can equate name spaces and IDs.
But it does require a fully hierarchical object space.

For the Chandra archive:
We get assigned, say, ID="SAO" for the name space of archives at SAO.
Then here is our ID hierarchy:

Object/namespace    ID

SAO                 SAO
Chandra archive     SAO:CDA
Chandra mission     SAO:CDA:CXO
Observation 2000    SAO:CDA:CXO:2000
Primary package     SAO:CDA:CXO:2000:primary
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

Since we are the issuing authority of the CXO name space, you can take
any two IDs that contain the substring CXO: and compare the
remainders.  As far into those remainders as they are identical there
is sameness, but you would probably have to look at the metadata to
determine the degree of sameness.  For instance, CXO:2000 in
NASA:HEASARC:CXO:2000 and SAO:CDA:CXO:2000 indicate that we are
pointing to the same observation

In the above example, the
remainders in the last three items are:
2000:primary:acisf02000N001_evt.fits
2000:primary:acisf02000N001_cntr_img2.fits
2000:primary:acisf02000N001_cntr_img2.jpg
The first and second are identical through 2000:primary:acisf02000N001_
Which means basically the same data and the same version (N001), but
in different data types (event list and image).
The second and third are identical through
2000:primary:acisf02000N001_cntr_img2.
which means (though it does not say that) that the jpg is derived from
the fits, representing the same image.

Now, in principle one could make sameness more explicit by changing:
SAO:CDA:CXO:2000:primary:acisf02000N001_cntr_img2.fits
to:
SAO:CDA:CXO:2000:primary:acis:N001:img:cntr:fits
Institution:archive:mission:obsid:datapackage:instrument:version:datatype:subtype:format
but that gets very detailed and, besides,
acisf02000N001_cntr_img2.fits happens to be a unique filename.
I have to admit, though, that it chopping the name up provides a more
general mechanism for judging sameness.


I admit that this is more of a use case than a requirements
discussion.  But I think it clarifies a number of things.
As I said at the beginning, I would suggest that:
1. We assume/require a one-to-one logical mapping between objects,
name spaces, and IDs
2. The object (ns, ID) space be strictly hierarchical
3. The issuer of objects in a given name space be responsible for
issuing authoritative IDs

The remaining question is whether there should be rules that govern
the hierarchy of IDs within an issuer's name space: should there be a
list of required ID elements (when applicable) such as instrument,
version, datatype, format in the example I gave.

  - Arnold

Ray Plante wrote:
> Hi Arnold,
> 
> Thanks for your comments.  I recognize that sameness will depend on (1) 
> the type of object be identified, and (2) the needs of a particular 
> application.  My answer to this was that (1) it would be up to the 
> curator to decide when two things are the same apart from its location, 
> and (2) standard interfaces should be specific about what kind of ID 
> should be given/returned.  
> 
> So, my question now is, what would suggest as an answer to how we use IDs?  
> Should we:
>   A.  Drop the concept of IDs altogether?
>   B.  Restrict IDs to refering to explicitly registered descriptions? 
> 	(Some other mechanism would be necessary for identifying different 
>          formats of the same image in the SIA image query results.)
>   C.  Go with a variation on the requirements and guidelines on "sameness" 
>         that I suggested?
>   D.  Something else entirely?
> 
> cheers,
> Ray
> 
--------------------------------------------------------------------------
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/
--------------------------------------------------------------------------