ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] Fwd: [New post] The Newest from SOA: The SOA Ontolog

To: "[ontolog-forum] " <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Christopher Menzel <cmenzel@xxxxxxxx>
Date: Fri, 24 Dec 2010 18:26:51 -0600
Message-id: <D0550E6D-EE77-4434-8EC1-7677CF90A43E@xxxxxxxx>
On Dec 24, 2010, at 4:49 PM, Ron Wheeler wrote:
> On 24/12/2010 2:54 PM, Christopher Menzel wrote:
>> On Dec 24, 2010, at 9:58 AM, Patrick Durusau wrote:
>>> ...My point is that useful ontologies are not limited to those constructed 
>based on formal logic and avoidance of modeling mistakes.
>> So, if I understand you, Patrick, your point is that
>> 
>>   (D) There can be useful models that are either (a) not based
>>       on formal language or (b) contain modeling mistakes.
>> 
>> Before addressing that, let's first be sure that we mean the same thing by 
>"based on formal logic."  By that I certainly don't mean an ontology that is 
>written in an explicitly formal language like CLIF.  There are, in particular, 
>some very nice, user-friendly graphical languages that are entirely rigorous.  
>To say that a language for constructing ontologies is based on formal logic is 
>only to say that the language has a sound logical foundation -- specifically, 
>the syntax of the language must be clearly defined (so that well-formed 
>constructs can be clearly distinguished from ill-formed) and the basic 
>constructs of the language must have a rigorous semantics so that assertions 
>in the language can be meaningfully interpreted. (Both are necessary, in 
>particular, if the ontology is going to be subject to any sort of useful 
>automated reasoning.)  A well-designed tool like Protege will simply 
>incorporate those logical foundations into the tool itself and a user will 
>adhere to them simply in virtue of using the tool.  So to say ontologies must 
>be "based on formal logic" is only to say that the representation language of 
>the ontology meets strike me as entirely minimal standards of rigor and 
>coherence.
>> 
> 
> It seems that you have just supported the original point by amplifying it.    (01)

Well, if by my having "supported the original point by amplifying it" you mean 
that I clarified it in such a way that any disagreement between me and Patrick 
(on this point) has been eliminated, then I'm happy to have done so.  Too 
often, those in the ontology community who lash out against "formal logic" — 
emphasis, typically, on "formal", often accompanied by a subtle sneer — are 
under the misimpression that those who tout its importance think that an 
ontology is legitimate only if it is written with lots of upside-down As, 
backwards Es, greek letters, and parentheses.  As I hope the above makes clear, 
the idea that ontologies be "based on formal logic" — emphasis on "based on" — 
is simply that the languages that ontologists use be well-defined and endowed 
with a rigorous semantics that affords clear definitions of standard logical 
notions like truth (under an interpretation of the language's chosen names, 
predicates, etc) and validity.  Only then can ontologies be shared without 
serious loss of meaning and reasoned upon reliably.    (02)

>> Given this, part (a) of your thesis (D) is that there are, or at least can 
>be, useful ontologies such that either it is not clear what is a well-formed 
>construct of the language and what isn't, or it is not clear what the meanings 
>of its basic constructs are.  I wouldn't want to go so far as to say such an 
>ontology could not be useful in some very limited way; perhaps in a small 
>organization in which people can clarify what the ontology is supposed to mean 
>with verbal explanations and where there is no need for the ontology to be 
>shared by others outside of that small setting.  So let me revise my claim to 
>be that ontologies "not based on formal logic" are useful at best in some very 
>limited setting.
>> 
>> As for part (b) of (D), I have to say I am nonplussed.  What do you have in 
>mind?  Obviously, it is unlikely that any large ontology with be completely 
>free of modeling errors, but surely the usefulness of an ontology is inversely 
>proportional to the extent of its errors.
>> 
> 
> Given the extensive discussion in this forum about the meaning of some very 
>basic terms and expressions of very common concepts (even ignoring the choice 
>of English words associated with them which causes all kinds of unintended 
>meanings drawn into terms), it is unlikely that any ontology will be devoid of 
>fragments that some very smart people who purport to be "Ontologists" find 
>defective or simply wrongheaded.    (03)

That could be true. Or not. Either way, this issue has nothing whatever to do 
with the points I've argued in this thread.  My point entails only that any 
usable ontology language must be such that there can be no reasonable 
disagreement about the meanings of its basic *logical* expressions or about 
such matters as how the meaning of a complex expression of the language is 
determined by the meanings of its simpler parts.  What you are talking about 
are the *non-logical* terms of an ontology, whose meanings might well be 
controversial and might well evolve over time.  A language "based on formal 
logic" in the sense at issue here can *help* with such problems by providing 
one with the means to clearly express what one intends that a given non-logical 
term mean by writing logically unambiguous axioms for it.  But having a sound 
logical foundation for one's ontology language alone can't solve such problems. 
 That comes only through careful empirical study, discussions with 
stakeholders, analysis by fellow ontologists, etc.  (That the advocates of 
sound logical foundations think that logic alone *can* solve such problems 
seems to be yet another annoyingly persistent myth floating freely about in 
certain quarters of the ontology community.)    (04)

> The usefulness of an Ontology will be judged by the ROI that it provides to 
>the user community which in turn will be determined by how it advances the 
>solution of a real world problem.    (05)

Sure thing.  Again, completely orthogonal to the issues I've raised.    (06)

> Most ontologies will likely be in a constant state of revision as new 
>concepts and relationships become known and need to be added to the ontology 
>to remove some defect or increase its scope. Think of the medical area or 
>financial services where advances and new products are daily occurrences.    (07)

And again.  But hey: hear hear.    (08)

> That does not mean that an ontology is not useful.    (09)

Sure thing -- so long as it is written in a language "based on formal logic". 
;-)    (010)

-chris    (011)


_________________________________________________________________
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
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (012)

<Prev in Thread] Current Thread [Next in Thread>