Re: [ontolog-forum] What goes into a Lexicon?

[William Frank <williamf.frank@xxxxxxxxx>]

I certainly agree with John Sowa's view that appropriate languages for ontologies must be rich enough to express what people know about their domains of expertise, rather than having to be poor enough to be machine decidable   I think this point is of vital  importance. I did not even know this was a matter of contention. 

Below I extract what I see as the backbone of his remarks, as they apply to OWL.    The history he recounts shows a pattern of thought I have seen often in technology. As a kind of architectural anti-pattern. I would call it the "Procrustian Bed": if it does not fit the technology, it must be re-shaped until it does.  Only then can we meaningfully talk about it. 

What is odd is that I have found this same attitude in the programmers in charge of application developments: "until you tell us the solution, (which one would think would be their job), telling us about the problem is not useful, and shows you are not among us experts."  And yet. I seem to find this same attitude often reflected in more august standards bodies.  Maybe this is just part of "information politics." 

We need to be able to express the *problem* (for example, describe the 
semantics of some domain of human endeavor) in a sufficiently rich 
language, even if a particular *solution* (such as being able to 
compute, for all cases, whether two specification of a 
semantics are equivelent) requires that we use a more restricted language.    If you only use the language of your particular solution, you remain locked in to that one solution, as has been the history of most software so far.  I imagined that this is the condition ontology work was intended to change.   Perhaps, by recognizing the different viewpoints of ontology specifiers and ontology implementers, we could steer clear of what looks like a dangerous turn of events. 

"

 Over time, however, several
proposed versions of logic, such as SWRL and RuleML, were
exiled from the layer cake because they were "undecidable".
Now, the only things left are RDFS, SPARQL, OWL, and RIF.

But the only implementation of RIF is the highly restricted
(and mostly unusable) subset that conforms to OWL semantics. 

The decidable fragment of OWL is
so restricted that anything beyond a toy example requires some
other language as a supplement

We need a bridge between those languages and the semantic system.
UML and related methodologies show that such bridges can be built,
and they have proved to be highly useful for mainstream IT.  But
you can't use a decidable language to define or specify what is
done with an undecidable language. 

 "

John

_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J

-- 
William Frank

413/376-8167


_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (01)