IVOA Vocabularies - WD

[Frederic V. Hessman]

Let me join the fray, though I haven't been getting all the emails  
for some reason.

First, the bottom line is the message, not the medium: we need a  
standard and non-optional form for IVOA vocabularies and a minimum  
common and standard vocabulary basis (SV).  We frankly don't care how  
we agree to organize vocabularies and define a standard basis, as  
long as we can agree on something soon.

Second, ...

> I appreciate the problem -- there are multiple astrophysical  
> vocabularies, as the document vividly points out.  I fail to  
> appreciate how adding a new vocabulary plus _two_ (count them) new  
> syntaxes addresses this.

See below.

>> Since there are multiple such standards,

> No, there's only one, RDF, plus a couple of standardised overlays.

The problem is, standards for what....?

>> they can be complex,

> RDF is about as complex as a brick.

Bricks are great tools for building houses, but a terrible means of  
creating shopping lists.   RDF is for probing the meaning of the web;  
I want an official but otherwise stupid list of astronomical terms.    
Yes, it's usually better to use the latest but tried-and-true nifty  
tools, it's usually better to be compatible with others outside our  
field, it's usually better to be able to do more rather than to do  
less, but it's always better to be able to do anything at all rather  
than wait for a committee to come up with the ultimate solution and  
do nothing.

>> One should also keep in mind that we might want to read this  
>> information
>> into astronomy client applications which just want the basic  
>> vocabulary,
>> do not need advanced inference capabilities, etc., and do not want to
>> be dependent upon some complex semantic web technology.

> If apps want to read, write and get value from RDF, with or without  
> inferencing, they can do so _today_ using industrial strength tools 
> [2]; if they want to read VOcabularies, they've got to wait for  
> someone to write, debug and port tools from scratch.

This sounds like an argument to stop the definition of simple XML  
documents entirely and start writing everything in RDF.  I hardly  
think that the task of taking a simple schema, converting it (semi- 
automatically) into useful code, and parsing a semi-colon separated  
list is the problem.  It's not as if people aren't dealing with  
simple schemata anyway.  Nevertheless...

> I and others have worn ourselves out trying to introduce RDF to the  
> VO (an area which is a more natural beneficiary of the RDF approach  
> than many areas which have applied it successfully).  I've tried  
> saying it's simple, and that it's sophisticated; I've tried saying  
> it's new, and that it's old; I've tried saying it's got pointy  
> brackets (just like XML, folks), and that it's readable; I've tried  
> using the word 'reasoning', and scrupulously avoiding it; I've  
> tried giving high-level arguments, and I have even tried, heaven  
> help me, showing running code; and ... nothing, no response, not  
> even much in the way of rebuttals, online or off.  No, the VO just  
> carries on inventing its own private syntaxes.  What is it I'm  
> doing wrong?

I can feel for you, Norman - the only reason I jumped into this was  
the exasperating experience of being unable to do something as  
trivial as _officially_ saying my image contains a star and a  
galaxy.   As usual, there are 1) those who want a brutally simple  
solution now, 2) those who want some semblance of the ultimate  
solution, and 3) those who would put up with a bit of effort to  
bridge the gap in between the two extremes.  I think I'm somewhere  
between 1) and 3).

I think everyone can agree that 1) a solution is needed, 2) the VO is  
unlikely to find/adopt a complex ontology-heavy solution any time  
soon, so 3) lets find a practicable compromise - now!.   VOcabulary  
and SV was our attempt at a compromise which preserved as much of the  
UCD-like legacy of the VO and used as little uncommon software  
infrastructure as possible.  You have to admit that the proposed  
VOcabulary itself  and the idea of the SV was REALLY simple and I  
assume most of Norman et al.'s wrath was directed at the attempt to  
sneak in ontological manipulations.  I will admit that our attempt at  
using UCD-like syntax in a primitive grammar was ..... let's say  
"daring": blame me for fantasizing that things might have been so  
simple and making the original suggestion.

> For what it's worth here's how I would address the problem of  
> multiple vocabularies:
>
>   1. For each of the vocabularies listed in the VOcabulary  
> document, describe a mechanical mapping from terms to URLs.  Use  
> the resulting strings as the interoperable labels (possibly  
> compacted, if you want, using namespace prefixes)
>
>   2. Use RDFS to annotate these strings nonexhaustively with  
> descriptions, aliases, and equivalences in other vocabularies.
>
>   3. Use standard libraries, in the language and on the platform of  
> your choice, to build on the result.
>
>   4. There is no step 4.
>
> This covers both the multiple vocabularies and the 'SV token  
> grammar'.  The only thing this would have difficulty with is 'NOT',  
> because it's well known that the concept 'not(X)' is hard to  
> discuss meaningfully in this context.  If you don't like that, then  
> there are other thoroughly shaken-down syntaxes from the last few  
> decades of computing which could be adapted to the purpose.

Well, not all of us are bricklayers, but I presume you mean something  
like:

	http://www.astro.physik.uni-goettingen.de/~hessman/rdf

which even includes my quick attempt at handling something like  
"NOT" ("notEquivalentTo").   I personally find all of this MUCH more  
complex and ill-constrained than a brutally simple XML document  
format where the schema gives me total control, but I suppose that's  
a question of taste....

In the meanwhile, let's assume that we COULD agree on the format of  
the VOcabulary's in SKOS/RDF: the real question is how everyone  
should use it  - the vast majority of VO-developers isn't going to  
want to jump out and start developing ontologies from the sudden  
hords of VOcabulary's, but to do simple things like,... well, tell  
someone there's a galaxy and a star in his/her image.  If we leave  
the UCD-syntax world (without abandoning the actual info) and  
actually use the RDF features we have been forced to create, then it  
looks like one has to do something like (here a VOEvent example,  
pretending that the RDF documents are all at IVOA):

	<VOEvent ...  xmlns:sv="http://www.ivoa.net/rdf/SV" ...>
		...
		<What>
			<Guess rdf:resource="&ucd;em.opt"/>
			<Guess rdf:resource="&sv;GRB"/>
			<Guess rdf:resource="&xyz;niftyIdea"/>
			...

Note that the RDF solution gives us connectivity to the web-world and  
visions of automatic ontologies, but robs us of the simple usage:

			<Guess ucd="em.opt;sv:GRB;xyz:niftyIdea">

which VOEvent types tend to like - about as succinct as it gets.  The  
avoidance of the terrible addtional task of dealing with semi-colon  
separated strings forces us in the direction of verbosity...... but -  
hey - as long as ya'll don't mind.  This is at least MUCH BETTER than  
random <Param>'s in the present VOEvent standard.

Please take a look at the link above and comment about/correct my  
bricklaying - If you can convince me that RDF bricks can be simply  
used as cuneiform tablets for vocabularies and that we can thereby  
gain moderately broad acceptance in the Semantic-VO community, I'm  
sold.   Think of it as a way of sneaking RDF into the VO, Norman.  We  
all promise not to use the word "ontology" and to keep chanting "This  
is simple.  This is the global standard.  These are just lists of  
tokens. ....".  Then our stab in the dark (or jab in the hornet's  
nest) won't have been in vain.

Rick

------------------------------------------------------------------------ 
------------------------
Dr. Frederic V. Hessman     Hessman at Astro.physik.Uni-Goettingen.DE
Institut für Astrophysik          Tel.  +49-551-39-5052
Friedrich-Hund-Platz 1         Fax +49-551-39-5043
37077 Goettingen                 Room F04-133
http://www.Astro.physik.Uni-Goettingen.de/~hessman
------------------------------------------------------------------------ 
-------------------------
MONET: a MOnitoring NEtwork of Telescopes
http://monet.Uni-Goettingen.de
------------------------------------------------------------------------ 
-------------------------


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.ivoa.net/pipermail/semantics/attachments/20070910/a5b6eba3/attachment-0001.html>