On Wed, April 27, 2011 11:14, Doug Skuce said:
> On Wed, Apr 27, 2011 at 8:47 AM, John F. Sowa <sowa@xxxxxxxxxxx> wrote:
>>
>> I'm starting a new thread because this subject cuts across numerous
>> threads, debates, disagreements, and confusions on this list.
>>
>> We have all agreed that the words and phrases of natural languages
>> are related to the formal definitions of an ontology. But many of
>> the disagreements arise from confusing one with the other. I'll
>> start with some observations (not exhaustive, but enough to start):
>>
>> 1. Whenever anybody talks about a formal definition in some version
>> of logic, the thing we're discussing is a formal ontology. (01)
> DS: Like Gruber's infamous definition, which included circuit diagrams,
> house plans, and Beethoven's 5th,
> this use of "thing" would cover, say, the formal definition of a group in
> math. But I would not call a group an ontology. But a group HAS an
> ontology (formal).
> Whenever people use the 'thing' word, watch out. It carrys no meaning,
> since "everything is a thing".
>
> This suggestion also implies there is no such thing as an informal
> ontology. But most so-called ontologies to date have been precisely that.
> So I say there are 2 disjoint kinds: formal (eg SUMO) and informal (eg
> Wordnet) (02)
The thing mentioned was a "formal definition", so sentence 1 could be
simplifed to "A formal definition in some version of logic is a formal
ontology." I would prefer to restrict the concept of ontology to a
description of some field of concern that specifies various classes of
thing, relations that may obtain among instances of such classes, and
relationships among such things. (03)
I would not consider WordNet to be an ontology, even an informal one. (04)
However, the disagreement over the meaning of terms should not be a
stumbling block. So long as we accept each others' defined classes as
valid, all that is needed is unique identifiers for them. (05)
>> 2. But whenever anyone says that a formal definition in logic is
>> not required, we're talking about a terminology. (06)
> DS No, we are talking informal ontology, PROVIDED the term is not isolated
> but is part of a VOCABULARY (all ontols provide a VOCAB) and some
> inter-term
> relationships, starting with narrower-than, are specified. Definitions can
> be in good NL, logic, or both. (07)
I would accept the idea of an informal ontology as being an ontology as i
defined it above, but without formal logic rules stating the
inter-relationships. (08)
I would not accept "narrower term" as an ontological relationship. (09)
> If you use the term `formal ontology` and the term `ontology` then either
> these are synonyms or there exists ěnformal ontology`.. Which do you
> intend?
>
> So here is my take:
> We must start with a VOCABULARY of 2 or more WORDS (which can be
> multi-words
> e.g. `formal ontology` is a multi-word. - there is no standard term for
> this). (010)
Wouldn't the standard term be "term" instead of WORD? And why not call
a multi-word term "multi-word term"? (011)
> If these are in a restricted domain (012)
Isn't any VOCABULARY in a restricted domain? (013)
> then terminologists call these TERMS.(I
> have worked closely with terminologists).Thus `formal ontology is a TERM.
> Probably 'ontology' is not, since it is in common vocabulary, at least for
> educated people.
> OTHERWISE they are just LEXICAL ITEMS (ie in any dictionary as a
> common-sense meaning).'ontology' is such, but it can also be a term when
> people like us use it in a restricted sense, like the term 'oblect' in
> programming.
> As soon as we specify inter-word relations we begin to form an ONTOLOGY. (014)
I would hold that a systematized set of inter-concept relations is needed.
It is important to distinguish between terms and the things they represent.
The ontology would concern the things represented, not the terms. (015)
> If these are specified using some form of logic,
> we are forming a formal ontol, else an informal ontol.
> Note that one could have a mixture: an ontol could
> be part formal and part informal. So you try to use logic, but when you
> are stuck you use NL.
>
> There is one most critical relationship: narrower-broader, or more
> precisely, subsumption or hyponomy. (016)
Go with the more precise concepts. A car engine is a narrower term than
a car, but it is not a hyponym. (017)
> Wordnet insists on this. (018)
> And a word may have more than one more general word: (019)
A concept referred to by a word may have more than one immediately
more general concept. I think you meant that. Let's make it clear
that an ontology is concerned with classes, relations, and instances,
not with words. (020)
Note that an ontology may have the concepts of Word, Term, EnglishWord,
MultiWordTerm, FinnishTerm, etc. Instances of these would be individuals
in the ontology. One word or term would not be more general than another. (021)
> restricting to single parenting does not wash in the real world. (022)
Agreed. (023)
> Lacking this relation, I would be loath to call
> a vocabulary an ontology.
>
> Ontologists would do well to study the Explanatory Combinatory Dictionary
> of Igor Melchuk. Such a dictionary is very much an ontology.
> http://olst.ling.umontreal.ca/melcuk/
>
>> 3. But sometimes, there are confusions about what kind of definition
>> is appropriate. Many things that look formal should actually be
>> called terminologies, and some terminologies can become formal
>> ontologies with minor additions -- for example, the terminology
>> for chemical compounds. (024)
>> 4. Things like WordNet and Roget's Thesaurus group words together
>> in "synsets" or clusters of closely related terms. But the
>> only definitions occur in the "glosses" or comments, which are
>> stated only in natural language. They are terminologies, not
>> ontologies. (025)
> DS They are informal ontologies. Why does everyone call them an
> ontology?
> There is a reason - everyone is not crazy.
> You are conflating the term 'ontology' with 'formal ontology' (026)
They are not ontologies as the word is standardly understood since
interrelationships among the synsets are not well defined. The
synsets do not have meanings, often terms in synsets have no overlapping
instances (such as male and female versions of some thing in the same
synset). (027)
Again, different people have different meanings for "ontology".
Philosophers don't accept any of the meanings that we are discussing. (028)
>> 5. Many terminologies about natural phenomena depend on cutting up
>> a continuous range of variation into a discrete set of categories.
>> Examples: river, stream, creek, brook, rivulet; tree, bush,
>> sapling; puddle, pond, lake, sea; ... Such terms cannot have
>> precise boundaries.
>>
>
> DS Yes, but what is your point? What is to be done with them? Cant just
> ignore a large chunk of language.
>
>
>> 6. Other terms depend on culture and technology, which are always
>> developing, mixing, and merging. Not so long ago, there was a
>> sharp distinction between a computer, a telephone, a television,
>> a typewriter, and a book. Today, the categories are blurred
>> and likely to swallow up other formerly disjoint categories.
>>
>
> DS Ditto
>
>>
>> 7. Finally, even when we have a formal ontology with detailed
>> axioms and definitions, the formal definitions will have
>> associated terms in natural languages -- examples include
>> the formal definitions in mathematics, science, and
>> engineering -- even some businesses, such as banking.
>>
>
> Of course there are terms - you cant have an ontology without a vocabuary.
>
>>
>> Much of the talk about interoperability can be confused by any
>> or all of the above issues. Some of the issues, such as the
>> APIs or the shape of a plug, require precise definitions.
>> But many of them depend on terminologies for which very loose
>> definitions are sufficient.
>>
>> It's also important to note that some differences, such as APIs
>> and plugs, can become interoperable by inserting an adapter, but
>> others can't.
>>
>> Suggestion: Whenever disagreements arise, ask whether the cause
>> of the disagreement is a confusion between a terminology or a
>> formal ontology.
>>
>
> DS You have been confusing 'ontology' with one of its two meanings. You
> are suggesting 'terminology' = 'informal ontology' (029)
I think not. I would consider 'informal ontology' to include far more
than a terminology and expect John Sowa would, too. (030)
> But a formal ontology HAS a vocabulary too. (031)
Agreed. (032)
> The term 'terminology' has a
> precise meaning when used by terminologists.
> Please check this out. We should get this clear, because this conflation
> of terms is at the root of the problem (033)
> - a *lexicon *is a collection of words it can also be
> called a *vocabulary *(eg: OED) (034)
Isn't a dictionary far more than just a collection of words or a
vocabulary? (035)
> - a *terminology *is a vocabulary used in a restricted sense
> in some domain.(eg mathematics, chemistry)
> - an *ontology *is a terminology given formal or informal
> relationship definitions, at least narrower-broader. (eg a KIF
> definition of some area in math. (036)
I would hold that far more is needed than narrower-broader. (037)
> - Note that math HAS a formal ontology without
> writing it in KIF. It is written in formal NL and can in principle be
> totally written in fol)
>
>
>
>
>> John Sowa
> --
> Doug Skuce PhD (Univ of Ottawa)
>
> 21 Torrington Pl
> Ottawa K1S 4E2
>
> 613 526 3732 (038)
=============================================================
doug foxvog doug@xxxxxxxxxx http://ProgressiveAustin.org (039)
"I speak as an American to the leaders of my own nation. The great
initiative in this war is ours. The initiative to stop it must be ours."
- Dr. Martin Luther King Jr.
============================================================= (040)
_________________________________________________________________
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 (041)
|