UCDs, metadata and AstroOntology

[Guy Rixon]

In response to Tony's paper on UCDs, I'd like to expand briefly on my view of
what UCDs are for.  For me, UCDs ARE metadata but ARE NOT names.  UCDs are
small collections of truths about (vectors of) data that allow a machine to
get a minimal understanding of how to treat those data.  Column names are
different.

I propose that a programme reading, say, a VOTable should be able to find and
use columns of that table for which it knows the semantics independently of
some ontological service.  I.e., the minimal metadata for using the table
should be included.  Example: find and plot all flux measurements in a
VOTable.  To do this, the programme could usefully search for an atom
"energy-flux".  It could also make use of a complementatry atom indicating
which filter system a flux came from; it might use this to lable a point in
the plot.  The programme would also need to know accurately how a photometric
measurement was presented; otherwise, it might try to combine incompatible
quantities on the same plot.  "Something to do with brightness" isn't a
sufficient description of flux-related quantiities.

Why UCDs as lengthy character strings?  Why not tuples, e.g. XML complex
types.

1. To make the parsing self-contained.  Collating bits of metadata from
many parts of a document makes a programme complicated.

2. To allow the UCDs to be read outside XML, where the only parsing is pattern
matching.

3. To keep the UCDs readable by humans.

Guy Rixon 				        gtr at ast.cam.ac.uk
Institute of Astronomy   	                Tel: +44-1223-337542
Madingley Road, Cambridge, UK, CB3 0HA		Fax: +44-1223-337523