UCDs vs ontologies?

[Rob Seaman]

On Jun 1, 2005, at 6:54 AM, Pierre Didelon wrote:

> But do/must UCD be the solution to (all?) natural langage parsing?

It would be premature to begin the work of the board by willy-nilly  
reassigning identifiers such as "em".  It seems likely that we will  
need to be able to express the concepts of both "electromagnetism"  
and "emission".  If there is emission, there is absorption.  And  
sometimes we may want to express the combination of "electromagnetic  
emission" at the same time.

On the other hand, we're told that the role of UCDs is distinct from  
that of ontologies.  An ontology is an (attempt) at expressing the  
complete range of some knowledge domain.  Astronomy is a big subject  
- its ontology will be big.  Perhaps by analogy we can view an  
ontology as the unabridged dictionary for some subject, whereas UCDs  
are simply one way to build a glossary for a specific purpose.   
Glossaries are often small enough to be appended to a brief document.

Personally, I think the VO community will need to develop several  
separate ontologies over time as well as several separate glossaries  
of UCDs or UCD-like constructs.  It is not obvious that a glossary of  
UCDs for tabular convenience is equivalent to a glossary of UCDs for  
VOEvent convenience.  An ontology can afford to be large and unwieldy  
to reach its goal of being complete and accurate.  A UCD style  
glossary, on the other hand, will eventually reach an optimum size.   
Its utility will pass a point of diminishing returns.  Too much  
precision engenders confusion.  The availability of too many options  
results in overlapping shades of meaning.

I gather the current list of UCDs was generated by looking at actual  
tables in the literature.  This is just how the unabridged OED was  
created from words sieved from millions of quotations.  Just like a  
dictionary, the work of maintaining the list of UCDs will continue  
indefinitely as new tabular usage is coined.

I would suggest that the creation of this new list of UCD-like  
entities to describe astronomical "concepts" is fundamentally a  
different exercise.  We may not be trying to generate a complete  
ontology with all interrelationships clearly drawn between all  
concepts, but we are trying to be complete in the sense of not  
leaving any gaps in the web of concepts.   "Star" and "galaxy" will  
clearly make the final cut.  "Star.white_dwarf" and "galaxy.spiral"  
most likely, too.  But it won't take many levels to exhaust the  
utility of compiling such a list.  I expect the final list to have  
hundreds of entries, not tens of thousands.

One final point.  The nature of this board is to participate in the  
process of certifying an official list of terms.  I think the true  
utility of both glossaries and dictionaries will be achieved when  
facilities are available for creating and maintaining *unofficial*  
lists.  For VOEvent, for instance, it seems likely that each project  
publishing events will adopt its own glossary pertinent to its own  
instrumentation and observations.  We should support these activities  
and provide a framework for project specific glossaries.  They will  
spring into existence whether or not we do so.  At least if we  
support the creation of project specific glossaries, we can have some  
say in controlling a common semantic structure and a standardized  
distribution mechanism.  This might also naturally lead to the next  
step of layering UCD glossaries on top of our emerging ontology 
(ies).  A glossary, after all, is nothing but a well chosen list of  
words out of the dictionary.  It is the dictionary that provides  
etymology, synonyms and antonyms, classification by part of speech,  
tenses and gender, pronunciation, ...

Sorry for the cross-posting.  If we can't restrain ourselves from  
generating all these mailing lists, I'm not sure what hope we have  
for a coherent set of UCD lists :-)

Rob Seaman
NOAO
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.ivoa.net/pipermail/ucd-tech/attachments/20050601/ccb525ef/attachment-0001.html>