How to choose?

[Rob Seaman]

On Sep 24, 2007, at 2:08 PM, Tony Linde wrote:

> Okay we seem to have equally good arguments from both sides of the  
> fence:
>
> a. develop vocab using SKOS and leave the ontology unless someone  
> else wants to do that in parallel, or
>
> b. develop vocab using OWL and derive SKOS version.

While I'm pleased with the traction this conversation is getting, I  
don't actually think we have coherent arguments yet for either side  
of the fence.  Robert Frost says:

	"Something there is that doesn't love a wall,"

But his neighbor (in "Mending Wall") says:

	"Good fences make good neighbors"

Google this and you'll see 1,120,000 hits, most relying on Frost's  
quote to assert things like:

	"Well-defined borders help prevent ethnic tension."

But the poet, of course, meant exactly the opposite:

	"'Why do they make good neighbors? Isn't it
	Where there are cows?
	But here there are no cows.
	Before I built a wall I'd ask to know
	What I was walling in or walling out,
	And to whom I was like to give offence.
	Something there is that doesn't love a wall,
	That wants it down.' I could say 'Elves' to him,
	But it's not elves exactly, and I'd rather
	He said it for himself."

Alan Weisman's excellent book, "The World Without Us", gives a sense  
for exactly how long all those walls would last without continual  
mending:  Not long.

> It seems that more people, of those who are knowledgeable and care,  
> prefer the SKOS approach.

Is this really an issue to take a poll on?  For astronomical  
purposes, consensus should be based on astronomical use cases.  Which  
of SKOS, OWL, something else (e.g., UCDs, SV as in the current draft,  
etc.) - or of natural language, the "none of the above" option - is  
best for VO purposes?

Is there even some reason to believe that all VO purposes (a pretty  
broad expanse of human intent) map onto the same semantic paradigm?   
Since much of the motivation for SV as we know it came from VOEvent,  
which of these options is best for the narrower VOEvent use cases?

> Can some of those ‘knowledgeable’ people get together this week and  
> really thrash out the options – maybe each person who favours one  
> side has to argue for the other side.

How about we all argue for the same side - for the efficient  
realization of useful astronomical goals?  Do we even care about the  
"best" solution?  Aren't we really engaged in an exercise of  
"satisficing" (http://utilitarianism.com/satisfice.htm)?

> Otherwise, about all I can suggest is that we leave the line open  
> to arguments for the rest of this week and then call for a vote  
> next week – unless some miracle occurs (http://news.bbc.co.uk/2/hi/ 
> uk_news/england/london/7010409.stm) and consensus breaks out.

What possibly would be the advantage of calling for a vote in the  
absence of consensus?  I don't know what passes for a miracle in the  
rest of the VO, but this really doesn't seem that intractable an  
issue compared to some others on which VOEvent has reached consensus.

I'm also a bit perplexed what exactly would be voted upon.  As we've  
been reminded, RDF, SKOS, OWL are already perfectly respectable  
standards. What precisely is keeping one VO project or another from  
using ontologies?  On the other hand, a standard IVOA vocabulary  
list, for instance, would be a data product held in common that  
requires sign off from the hierarchy.  An assertion that "we don't  
need no stinkin' vocabulary" does not imply "we need to pick between  
SKOS and OWL".

A few questions:

- What happens when the ontology evolves?  After all, we're talking  
about methods for discovering and characterizing brand new  
phenomena.  (http://www.urticator.net/essay/2/217.html)

- How efficient are the tools of different kinds for applications  
with split second timing requirements?  (http://alphard.ethz.ch/ 
gerber/approx/default.html)

- How robust are they against network or server outages?  KPNO has a  
policy against mountain data-taking systems depending on network  
resources for critical functionality.  Presumably the corresponding  
policies for robotic networks are even more strict.

- What can SKOS and OWL do for simple use cases?  For example:

	1) A high signal-to-noise optical transient is detected.

	2) A VOEvent packet is generated.

	3) Its discovery "signature" is compatible with a wide range of  
possible underlying phenomena.

	4) Its brightness, location, rarity, etc., make it of interest to  
several subscribers.

	5a) How do semantic technologies aid in the efficient and reliable  
characterization of the phenomenon?  (http:// 
oaei.ontologymatching.org/2007)

	5b) What strategy is best used by the several subscribers to work  
together compiling follow-up observations for the common good?   
(http://www.iscid.org/encyclopedia/Nash_Equilibrium)

- Would SKOS or OWL (or SV, for that matter) perform better than the  
status quo (natural language)?  (http://en.wikipedia.org/wiki/ 
Impedance_matching)

- If this is not a well posed use case for differentiating between  
our semantic software choices, then what would be?  (http:// 
www.econlib.org/Library/Enc/TragedyoftheCommons.html)

- Rob