Alberto Micol wrote:
>
> Dear All,
>
> I'm trying to catch up with UCDs, sorry to be so late.
>
> Matteo's and Ana's comments, plus my own way of reading the list of words,
> and especially the comments associated with each word, make me think that
> something is not yet properly defined, and that some ambiguity is still
> to be removed.
>
> While historically UCDs were assigned to *columns* in the Vizier
> collection of tables,
> I wonder if this is still true. That is, I cannot find in any document
> of the UCD1+
> the answer to the question:
>
> What is that a UCD can be assigned to?
>
> I know that it CAN be assigned to a scalar quantity;
> I know that it CAN be assigned to a column in a table;
>
> but
>
> CAN it be assigned to an image, to a table, or any other piece of
> Astronomical Data?
>
> I cannot find the answer in any recent UCD-related document (correct me
> if wrong).
> I think we need to fill this gap, and give a clear answer.
>
> The answer can be:
>
> NO
> ---
> If the answer is that a UCD can NOT be assigned to anything else than a
> scalar quantity
> or an array of identical quantities (i.e., the column in a table),
> then I do not see why we should have a UCD like instr.interferometry
> (Thanks Matteo for the good example)
>
>
> YES
> ---
> If, at the contrary, UCDs can be assigned to generic, however
> complicated, data structures
> then instr.interferometry becomes a nice UCD (though I would suggest
> data.interferometry instead).
>
> Furthermore, if UCDs can be assigned to complex data structures then
> suddenly I can make sense of other UCDs (like instr.filter.transm)
> which are currently defined in an ambiguous way:
>
> > instr.filter.transm | Filter transmission
>
> Is it for a typical value of the transmission of my filter (e.g. 0.2)?
> or
> Does it refer to the entire array that represents the "transmission
> curve" ?
May be it can be both, depending of the context and the data decribed?
But transmission curve is certainly more precisely described by SED-DM,
and in this context why not associating this structure with such an UCD?
>
> On the other hand, if the answer is YES, then a table could have its own
> UCDs,
> (and in that case the VOTable standard will need a revision to support
> UCDs at the table level).
It is the only "special" case where ucd are allowed aside column (FIELD) or
"individual" (which can contain an array nevertheless) quantity (PARAM),
not speaking of column or quantity grouping (GROUP) ;-)
>
>
> But I think that the answer is NO, because otherwise the vocabulary
> would need to explode
> and include all kind of things.
> Hence, if the answer is NO, then I would suggest to make the comments a
> bit more strict
> to avoid confusion.
It is perhaps neither YES nor NO.
Actually the main usage is still column characterization
but who can forsee all possible usefull usage... and/or try to forbid it?
> A couple of examples:
>
> Current Description Possibly less ambiguous description
> ------------------- -----------------------------------------
> Filter transmission -> Typical value for the filter transmission
> Spectral resolution -> Typical value of the spectral resolution
> Resolution -> Typical value of the angular resolution
> Spectral line -> Common name of the spectral line
> Detector -> Name of the detector
> Photographic Plate -> Type of emulsion
> etc. etc.
>
> Alberto
Friendly, Pierre Received on 2005-07-19Z14:39:07