Re: [ontolog-forum] Architectural considerations in Ontology Development

[Amanda Vizedom <amanda.vizedom@xxxxxxxxx>]

On Tue, Feb 19, 2013 at 6:39 PM, John F Sowa <sowa@xxxxxxxxxxx> wrote:

Ed,

The following definition is about as neutral and non-controversial as
you can get.  All other definitions of 'ontology' are special cases.

JFS, as derived from definitions in the Free Merriam-Webster online:

>> a particular analysis of a set of facts in their relation to one another
>> about the nature of being or the kinds of things that have existence

EJB

> If this were the definition we use in the Ontolog Forum, I would
> unsubscribe immediately.

I cannot see why you or anybody else would object to it as a general
statement that subsumes all the definitions anyone has proposed.

JFS

>> To specialize this definition for computer applications, delete the
>> option "the nature of being" and add "in some domain of application":
>>> a particular analysis of a set of facts in their relation to one another
>>> about the kinds of things that have existence in some domain of application

EJB

> So now we are not using the dictionary definition, and we are assigning
> a very different semantic interpretation to "kinds of things that have existence".

This is just a specialization of the general definition.  There are
only two differences:  (1) delete the option "the nature of existence",
which is more philosophical than computational, and (2) add the point
that it is specialized for a particular application domain.

In this forum, we have been talking about ontologies for particular
domains for years.  I don't know what you object to.

JFS

>> Note that neither the OED nor Merriam-Webster's nor the long tradition
>> in philosophy require theories to be stated in formal logic.

EJB

> So, the idea that "computational ontology" is based on "formal logic"
> is not a part of your conceptualization of "computational ontologies".

This debate is typical of what happens when you focus your energy getting people to agree on NL terms, especially when their backgrounds and current contexts of use do not match extremely closely.

Fortunately, it is also unnecessary. We do not need to agree on the meaning of the word "ontology."  This only need to agree, to a certain extent, about the topic and scope of the conversation here. We can have useful conversations so long as we do that, regardless of what we think about other uses of the word or about other matters beyond that scope.

It is useless to argue about a general, all-purpose definition of 'ontology.' That is a natural language _expression_, and like so many others, it has a variety of different meanings (In WordNet, it will have multiple word senses, each of which belongs to a different Synset. In a formal, computational ontology with lexical mappings, it can properly be specified as a label, or "use for" string for multiple concepts, including, but not limited to, the technical concepts it expresses in philosophy and computer science, respectively, and to looser, non-technical concepts in lay use, such as those in general-purpose dictionaries. [And by the way, as someone trained and experienced in both computational and philosophical ontology, I do not think that one specializes the other, or that both are specializations of some more general, lay "ontology". The two technical fields have elements in common, but they bear a family resemblance relationship, not a subsumption or direct-common-ancestor one.]

It is immensely frustrating, and a repeating straw on the back of the camel that takes people off of this list and out of this community, when conversation comes back to this type of red herring again and again.  

Participating in a social discussion forum involves learning and abiding by the agreements of that forum, at least during the course of ordinary exchange. There is plenty to discuss and debate such a framework. This list is a forum of the Ontolog community of practice. The Ontolog Charter (http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologCharter) does include "both formal and informal ontologies." Unless or until that changes, we who participate should respect that both are in-bounds. The Ontolog Charter also makes clear that this community of practice has a orientation toward "practical issues and strategies," "development and application," "ontological engineering" (and by implication, the sorts of ontologies that can be meaningfully engineered), and "semantic technologies."  These are concerns associated with ontologies as understood by computer science and information science, for example, and not the philosophical or lay concepts. And for many participants, that charter, as minimal as it may seem, is a condition of our participation. The more frequently or longer the conversation wanders away from this focus and toward philosophical or very general concepts of ontology, the more steeply morale and desire to participate decline. 

I've been teaching logic for many years, and I know very well that(
most programmers don't know how to express themselves in logic.

This is true, IME, but irrelevant, in that most programmers aren't ontologists. Formal ontologists do need to understand formal logic, insofar as they need to understand the specified semantics and implications of the language in which they develop. Informal ontologists need to understand informal logic to the extent that they use logical notions such as material implication, conjunction, disjunction, negation, all, some, etc., to express ontologies; they need to understand the implicit entailments of what they render explicit, if their ontologies are to be understandable to others.

 

In fact, Doug Lenat said that his standard practice is to give
all prospective hires a short exam of a few sentences in English,
which they have to translate to FOL.  There are no ambiguous or
tricky sentences, and they don't go beyond FOL.

I don't know whether Doug Lenat ever said that, but it is false. What's true is that the formalization doesn't matter; someone could do well on the interview tests who has never used *symbolic* logic. It is not a matter of "translating" between English sentences and FOL expressions. Rather, it is a matter of capturing "common" or "simple" facts, expressed and English but carrying their full ordinary assumptions and interpretations, in some logic-based representation.  In order to succeed, an interviewee has to have very good logical intuitions, though indeed they may not have formal logic training, and they must use those intuitions to think through implications quickly. There are always tricky bits, where the logic newbie is likely to go for a simple representation that has undesired implications, or omits essential ones.  A logic newbie with strong intuitions might make such mistakes, the question then is what happens when the undesired or omitted implications are pointed out. Does he or she quickly understand the error? Can he or she revise or add to their representation to correct it? If so, they can learn FOL, CycL, or other formalisms. On the other hand, folks can be trained in one or more FOL representation and never get the hang of representing things in it. So that informal test is a pretty good predictor of potential as a formal ontologist. This is true even though if the candidate succeeds in getting far enough into the representation, it will, in fact, go beyond FOL.

Best,

Amanda

_________________________________________________________________
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)