ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] Levels

To: "[ontolog-forum] " <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Kathryn Blackmond Laskey <klaskey@xxxxxxx>
Date: Sun, 11 Feb 2007 19:26:03 -0500
Message-id: <p0611040cc1f535bf1fc8@[192.168.0.103]>
John,    (01)

>As I have said many times, I have no objection whatever
>to saying that microtheories are part of the ontology.    (02)

To clarify, by "the ontology," you emphatically do NOT mean the same 
thing as what your previous email called the "base ontology (i.e., 
what is usually called the 'upper level').    (03)

You are distinguishing among several levels:
  - Upper ontologies such as sets, arithmetic, space and time, ...
  - A base ontology for a domain such as medicine.
  - Microtheories for particular sub-domains, such as theories
    about particular diseases. You have said you're willing to
    call these "extended ontologies".    (04)

Am I tracking you?    (05)

Practitioners use the word "ontology" for all three, and may not 
differentiate among levels.  This creates some of the very problems 
ontologies are supposed to solve.  For example, we don't want a 
medical ontology contains highly specialized axioms for calculating 
dosages of drugs from patients' weights, which turn out on careful 
analysis to be reinventions of arithmetic.  It would be much better 
to have the medical ontology reuse a standard ontology for arithmetic.    (06)

I'm in agreement that we need different levels.  Some purists argue 
that only the uppermost level deserves the name ontology.  But the 
engineering community seems to have co-opted the term and given it a 
much broader meaning.  It's probably not possible to close Pandora's 
box now.  So we need to differentiate the levels.    (07)

Should we go for "upper ontology", "base domain ontology", and 
"extended domain ontology"?    (08)

These questions are entirely orthogonal to the question of whether 
ontologies should be formalized in classical logic, probabilistic 
logic, or some appropriate combination for the purpose.    (09)

>  > SUMO hasn't been officially adopted yet, has it?
>
>I don't know what you mean by "adopted".  A lot of people
>use it.  I think that SUMO, Cyc, and many other ontologies
>are quite good for their purposes.
>
>But the fact that there are multiple widely used ontologies...    (010)

That's what I meant.  There are multiple widely used ontologies even 
at the "upper" level where you said all reasonable people were 
supposed to agree.    (011)

The point I was making was was that reasonable people don't agree.    (012)

>implies that any common upper level should contain only
>the intersection, not the union of them.    (013)

Is the intersection non-empty?    (014)

>  > So you don't want a medical ontology to represent that
>  > medical tests have sensitivity and specificity?
>
>I wouldn't prohibit any kind of information from a domain
>ontology.  If sensitivity and specificity are conventional
>terms (i.e., at the level of planet vs. dwarf planet), then
>they definitely belong in a very general ontology for medicine.
>But the detailed values that have been obtained by observation
>belong in more detailed subdomains.    (015)

Agreed.    (016)

>  > In many such cases, it would be more accurate from a medical
>  > standpoint to define the condition by means of a probability
>  > model in which presence of the condition is a hidden
>  > unobservable variable whose probability is inferred from
>  > observable indicators.  That would require probabilistic
>  > axioms in the ontology or microtheory or extended ontology
>  > or whatever we are going to call it.
>
>I would agree that the idea of probabilistic models deserves
>to be specified at a very high level and that specific models
>of that kind should be availabe in domain ontologies at various
>levels.  But a particular theory of a particular illness would
>be a very specialized model that would be better suited to a
>lower-level domain ontology.    (017)

Agreed.    (018)

>For example, smallpox has been diagnosed with considerable
>accuracy for centuries, but the medical understanding of
>the disease has changed enormously.  In order to relate
>medical data from different periods of time, it's necessary
>to have underspecified definitions at the higher levels, but
>more detailed specifications, which may change over time,
>at the lower levels.    (019)

You put the status of planets at above the level of particular 
diseases -- but Pluto has just been reclassified!    (020)

>The definition I proposed on January 30 includes everything
>you've requested so far:
>
>      A formal ontology consists of a theory T stated in some
>      version of logic and a nonempty vocabulary V.  The
>      vocabulary V is a subset of the names of types and relations
>      used in T.    (021)

If we allow "some version of logic" to include a probabilistic logic, 
I'll go with it as a necessary condition for being an ontology.    (022)

But allowing your "loopy letter" example below to qualify as a bona 
fide ontology seems to go against the spirit of what I thought 
ontologies were supposed to be, unless you could demonstrate to me 
that an ontology of loopy letters was truly useful in some domain.    (023)

>Following is a better specification of an ontology.
>
>Vocabulary: {loopyLetter}
>
>Theory:
>
>     Every loopyLetter is a letter in a circular envelope.
>
>     No loopyLetter is delivered on a Tuesday.
>
>This would be a special-case ontology that uses terms that may
>be defined in other ontologies: letter, in, circular, envelope,
>deliver, Tuesday.    (024)

Wouldn't you want to require your ontology to explicitly name the 
ontologies from which it was drawing terms?  Otherwise, we could have 
any number of meanings assigned to the terms not defined in the 
ontology, if there were multiple external ontologies that defined 
them in conflicting ways.    (025)

Kathy    (026)

_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Subscribe/Config: 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 Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (027)

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