[Voevent-core] Fwd: standard vocabulary

[Rob Seaman]

Hi Francois,

> In my mind, it's (hopefully) more than just flexibility -- it is
> a description of the phenomenon. Assigning a name to it does not
> describe it -- you could name it 054321ff instead of GRB and the
> semantic contents would be the same...

I think you have put your finger on it.  First, note that I was not  
questioning the good work that has gone into UCDs for their original  
purpose.  In the case of describing a keyword or a column of a table,  
the point is that the column already has a name attached, whether it  
is "RA" or "EXPOSURE" or whatever.  The UCD in this case is providing  
additional context to understand that "MAG_V" and "V_MAGNITUDE" are  
both expressions of a quantity representing "phot.mag;em.opt.V".  The  
UCD unites the disparate quantities into a single semantically  
manageable notion.

But "GRBness" is not a scalar with some floating point value  
attached, it is itself a state to be named.

> The "atoms" used in
> "process.variation.burst;em.X-ray" have all a definition in the
> UCD dictionary, their association therefore means something.

But what they mean is not intrinsic to the astronomy.  The definition  
is not fixed in advance - it is the point of the whole exercise.  A  
GRB or AGN or SN is a complex object/process with many faces that  
depend on how it is observed and from what preferred direction.  A  
GRB (or a class of same) may result in several distinct phenomena  
that astronomers are seeking to understand and unite into a single  
coherent entity.  The atoms in this case are dividing the science,  
rather than uniting it.

> Of course it would be possible to add "GRB" among the atoms,
> but then you remove the possible relationship between e.g. GRB
> and X-ray burst.

But that is purely a phenomenological relationship, whereas there are  
many actual relationships of cause and effect through complex physics  
that these UCD-like expressions do nothing to address.  A supernova  
(of a particular type) is the result of a binary star system  
exchanging mass that eventually triggers TNR.  A nova is some other  
binary star system whose mass exchange triggers only surface  
burning.  There are similarities - and there are differences.   
Additionally, the notion of evolution is missing - one class of  
object may become another class of object.  I wouldn't expect UCDs to  
help much here, but they shouldn't hinder by implying that a GRB "is"  
a "process.variation.burst; em.gamma".

Which is to say that an gamma-ray burst observed by Swift is used to  
infer the existence of something we call a "GRB" which exhibits many  
other observable aspects.  On the other hand, the UCD  
"process.variation.burst; em.gamma" explicitly misses those more  
physical implications.

> In other terms, the UCD is a way to describe
> phenomenae, observables, parameters, etc... with a restricted set
> of well-defined "words".

Will agree with parameters and possibly observables.  Don't think  
that we've capture "phenomena" yet.  In the latter case, the  
vocabulary comes after the fact, not before.

Rob