VOConcepts paper

[Hessman Frederic]

> Please find attached a white paper from Andrea Preite-Martinez, the  
> chairman of the IVOA Semantics working group, which is a first  
> attempt to build a standard vocabulary for astronomical phenomena  
> and for astronomical objects. I would like to have a brief  
> discussion of this at the VOEvent meeting next week, and/or receive  
> comments by email from those who cannot attend the meeting. Rick  
> Hessman has already worked on exactly this topic.
>
> I would like to compile a "group response feedback" on this paper  
> for Andrea and the Semantics WG.
>
> Roy Williams

Just a note as to the origin of the proposal:  it's not something out  
of the blue but basically my original VOConcept ideas and vocabulary  
modified slightly, extended and sometimes much improved by Andrea et  
al.  The two of us have had a few email exchanges about what we think  
the syntax should look like, and we basically were able to agree on  
most but not all things.  I do have a very few comments/changes I'd  
like to see made to this version:

- The list needs to be reduced to the atoms and then have a list of  
typical/interesting/expected examples afterwards - the current
	list combines both.  The examples are good to see just how  
successful the exercise has been, but the basic discussion needs
	to be focused on the atoms.

- I'm not real crazy about em.VeryRed or source.VeryRed since "red"  
doesn't mean much all alone and what's the difference 		
	between "red", "VeryRed" and "ExtremeRed" (sic)?  How about just  
something like source.red;em.IR or even
	source;em.IR;stat.extreme?  Tricky, since is the emission or the  
source or both "extreme"?  Or do we start getting complicated
	and use parentheses:  source;(em.IR;stat.extreme) ???  In this case,  
I'd say "red" is good enough!

- class.betaLyr, not class.betLyr   etc.    We shouldn't bother to  
save 8 bits.

- A pundit's complaint:
	stars.variable.nova -> stars.variable.cataclysmic.nova (or add  
process.explosion)
	stars.variable.nova-like -> stars.variable.cataclysmic;class.novalike
	stars.variable.dwarfNova -> stars.variable.cataclysmic;class.dwarfNova

- stat.partof -> stat.partOf

- ISM.Herbig-Haro -> ISM.nebula.Herbig-Haro  or   
ISM.nebula;class.Herbig-Haro?    or    ISM.nebula;process.jet?

- stars.variable.cepheid but stars.variable;class.RRLyr ???  How  
about stars.variable;class.deltaCep ???

	This example shows how difficult it is to know just when to express  
atoms or to create combinations!  Another good example
	is "stars;class.Be" and "stars.AGB" instead of "stars.Be" or  
"stars;class.AGB"?   We should have a clear-cut
	system for deciding what needs to be described as a "class.*" and  
what directly as a sub-atom.  How about the distinction 	
	between physics and phenomenology?  Difficult, since we'd hope that  
phenomenology would quickly change into physical
	explanations.   The class atom really has the function of a name (or  
an XML attribute) since Andrea has made it a giant
	grab-bag of things, which gives it a very unusual position in the  
world of UCD's.    I personally prefer the purely hierarchical
	version (e.g. stars.Be), even if it means that some of the concepts  
start to get out of hand (e.g.
	stars.variable.cataclysmic.dwarfNova.UGem).

	Andreas intent as listed on the first page - to use "class" when  
there's an architype involved (e.g. "RRLyr") is fine, but needs
	to be done consistently, e.g.
		stars;class.Be	->	stars.Be
		stars.cepheid	->	stars.variable;class.deltaCep (and not with mixed  
forms like on page 4)
		stars;class.Wolf-Rayet	-> stars.Wolf-Rayet
		stars.superNova;class.Ia	->	stars.supernova.typeIa (there is no  
"Ia" object)
		galaxies;class.HII 	->	galaxies.HII (there is no "HII" object)
		galaxies.AGN;class.Blazar	-> galaxies.AGN.blazar (whereas  
galaxies.AGN;class.BLLac is OK)
	and whether an object class is named by an archtype or not is pretty  
random, so I think this class.* form is really unnecessary
	for all but purely semantic history reasons.

- galaxies.AbsLineSystem -> galaxies.absLineSystem   or   
galaxies;phys.absorption;stat.multiple??

	Never begin a sub-atom with a capital letter unless needed (e.g.  
proper names like Herbig-Haro).

	Since we could just as easily have ISM.AbsLineSystem, when does one  
use the same sub-atom and when does one use
	combinations?

VO-StandardVocabulary or VOConcepts (whatever you want to call it)  
are a true Pandora's box of ontological cunundrums, but we
have to start somewhere!   I'm sure both Andrea and myself would like  
hearing your comments!

Rick
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.ivoa.net/pipermail/voevent/attachments/20051130/1c3c5d9d/attachment-0001.html>