[Top] [All Lists]

Re: [ontolog-forum] CL, CG, IKL and the relationship between symbols in

To: Duane Nickull <dnickull@xxxxxxxxx>
Cc: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Pat Hayes <phayes@xxxxxxx>
Date: Thu, 3 Jan 2008 11:02:52 -0800
Message-id: <p06230907c3a2d7d7e53c@[]>
>Answers inline:
>>>  Context is a set
>>  You are speaking in the singular. Do you mean to
>>  imply that that is a single thing called
>>  "context" which is a *particular* set?
>DN: The manner in which CCTS did this is to establish there were 8 major
>factors that influenced the manner in which a generalized data element
>concept becomes a very specific data element.  This was referred to as the
>context which implied a set of things.  It was deemed non-exclusive as there
>was no way to know if it was complete.
>>>  of zero or more qualifiers
>>  What is a 'qualifier'?
>DN: It is a UN/CEFACT term.  It roughly translates to an external force or
>asepct of context    (01)

Wait. You are here defining "context". You cannot appeal to an 
'aspect of context' in the very definition of context.    (02)

>which has a specialization effect on the instance being
>discussed.    (03)

Oh, well then that is far too limiting as part of a general 
definition, as contexts can do far more than this. A conversational 
context, for example, can introduce entities into the common ground 
that were not present before. In Cyc's 'microtheory' view of 
contexts, a context can change the meaning of a word in many ways, 
not always restrictive.    (04)

>  Example:  If I am an instance of the human class (some say this
>might be a stretch), and was in Canada, this makes me a Canadian human,
>being more specific than a mere human.
>>>  that affect various aspects of
>>>  the semantics of a given statement.
>>  Of the semantics or of the meaning?
>DN: Specifically in UN/CEFACT terms, it modifies the semantics by making the
>instance more specialized for the specific context.  A generic date data
>element can be refined and constrained to be a "PurchaseOrder.Ship.Date"
>which is a type of date that takes on additional characteristics.
>Note that all of the preceding are done during the modeling work to refine
>metadata by making the semantics of data elements more specialized.    (05)

Um... I think we are at cross purposes about what "semantics" means. 
In my terms, none of this changes the semantics (in fact, it would be 
part of the semantics). But OK, I get the idea.    (06)

>  >
>>>   Different context qualifiers can impact
>>>  one or more aspects of the semantics of such statements including
>>>  representation terms and concepts.
>>>  Example:
>>>  A glass of water
>>  What is this an example of? Is the glass of water
>>  the topic of an ontology, or an example of a
>>  representation of something?
>DN: An instance of a class of either a taxonomy of ontology.    (07)

OK, I see. The class of glasses of water. But one cannot see a class, 
or have it sitting on one's kitchen table.    (08)

>  Perhaps this
>is not the best example as I noted after I wrote it.
>>>  Context one:
>>>  Glass of water is sitting on your kitchen table.  To the average Western
>>>  observer, it is quite in it's place in this context and is not raised in
>>>  one's internal tuple stores.
>>  ?? What internal tuple stores are you talking
>>  about? I very much doubt if my brain works using
>>  RDF. And what does 'raise' mean?
>DN: an analogy to say that you would not raise it in your general level of
>awareness in this context as there is nothing that strikes you as unusual or
>out of place.  OTOH, if you were very thirsty, you may immediately fixate
>your attention on this object.    (09)

OK, but what has all this got to do with what (I thought) we were 
talking about?    (010)

BTW, you might find this interesting, its a critical survey I wrote 
for one of the context workshops a few years ago, and makes a few 
distinctions which are relevant here.    (011)

http://www.ihmc.us/users/phayes/Pub/ContextsInContext.pdf    (012)

>  >> Concept:  water, two molecules of hydrogen bound to one of 
>oxygen in a glass
>>>  container.
>>>  Dangerous: possibly
>>>  State: liquid
>>>  Use: quenching thirst
>>>  Mass: about 200 grams
>>>  Context two:
>  >>
>>>  You are driving down the road in sub zero weather and the same glass of
>>>  water is in the middle of the road.
>>  No, it isn't. Not the SAME glass of water. OK,
>>  these are two different circumstances, and if
>>  fully described in an ontological formalism would
>>  have very different descriptions. But what his
>>  this fact to do with contexts? Ontologies aren't
>>  living in the world like we are, driving cars and
>>  drinking glasses of water. Living in a changing
>>  world raises a host of new issues that go beyond
>>  ontology engineering. This kind of difference you
>>  describe here is more relevant to KR in AI than
>>  to ontology engineering.
>DN: Perhaps.  I am here to try and learn/understand this.  Your help has
>been great so far.
>>  Here's a quick way to say the difference between
>>  OE and AI. In many AI applications, a reasoner is
>>  living IN a changing world, one that it needs to
>>  perceive and act in. In OE applications, the job
>>  of the ontology is to DESCRIBE possibly-changing
>>  worlds, but not to live IN them.
>DN: makes sense.
>>>    Since it represents a hazard for your
>>>  vehicle, you raise it to the highest layers of consciousness in your
>>>  internal tuple stores and give it active attention.
>>  Ah, this is what 'raise' means? OK, but what has
>>  this to do with ontologies? AFAIK, the notion of
>>  attention simply doesn't arise in mechanical
>>  reasoning.
>DN: it is sort of orthogonal to ontology however in the CI field    (013)

CI?    (014)

>, cognitive
>skills rely on an agent to recognize that X is an instance of
><ontology_class> or <taxonomy_item> etc.  The level of attention a given
>instance X has in any observers consciousness (or machine equivalent tuple
>stores), varies depending on the surrounding context in which the instance X
>incurs.    (015)

Well, I don't know of any tuple-based mechanisms which use any notion 
like this of raising attention levels. It might work, would be 
interesting project to try. I guess Im not sure what would count as 
the 'surrounding' context of a tuple store or an RDF reasoner, though.    (016)

>  This is of course not a 100% proven hypothesis and only my opinion.
>I am not sure if others agree.    (017)

IHMC            (850)434 8903 or (650)494 3973   home
40 South Alcaniz St.    (850)202 4416   office
Pensacola                       (850)202 4440   fax
FL 32502                        (850)291 0667    cell
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes    (018)

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    (019)

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