Sean and Pat, (01)
You both have valid points. To clarify the issues, I suggest
that we adopt some terms that are common in discussions of
meaning in languages, both natural and artificial: (02)
- *Syntax* addresses the grammatical form of any notation,
independent of any question of what those forms mean. (03)
- *Semantics* addresses the relationship between any
notation and the things or events that the notation
refers to. It includes both reference (what the names
and variables refer to in the domain of interest) and
truth (whether the patterns expressed by the notation
correctly characterize the patterns in the domain). (04)
- *Pragmatics* addresses the purpose of the notation
and how it relates to the behavior of the people and
other agents (including computers) that use the notation. (05)
SB>> The meaning of a computer system is always the behaviour
>> of the organization that uses it. (06)
PH> ... that is rather a stretch. The meaning IS the behavior??
> No, the behavior depends (in part) on the meaning: but the
> meaning is what it is even if nobody acts on the information. (07)
That is why the word 'meaning' is confusing. Sean is talking
about pragmatics, and Pat is talking about semantics. I suggest
that we avoid the word 'meaning' and discuss the issues in terms
of syntax, semantics, and pragmatics. (08)
SB>> If I have a collection of designs for an aircraft, a set of
>> orders for parts and a stock of material, I expect to come in
>> the following day to have a set of appropriately shaped lumps
>> of metal, not a print out saying 'the machine could in theory
>> make the parts you require'. (09)
That is a valid goal. But it is a matter of pragmatics (what
the system is supposed to do), which depends critically on
the semantics (what are the specified patterns for the lumps)
and the syntax (what is the grammar for specifying lumps). (010)
PH> I doubt if it is possible to quantify this or even define
> it reasonably, as it will depend so heavily on what use is
> made of the information in the ontology. (011)
My revision of Pat's point: (012)
I doubt if it is possible to quantify [the pragmatics] or
even define it reasonably, as it will depend so heavily on
what use is made of the [semantic] information in the ontology. (013)
I certainly agree that the pragmatics depends on the semantics,
but I also believe that it is possible to define procedures that
would determine the pragmatics -- i.e., how a computerized machine
should respond to the semantics by making the desired products. (014)
SB>> Perhaps a more useful question is not whether the ontology
>> corresponds (or what ever phrase does not upset the philosophers)
>> to reality, but what is the risk that I run if I assume you mean
>> the same thing that I do when we use a term from an ontology. (015)
PH> I doubt if it is possible to quantify this or even define it
> reasonably, as it will depend so heavily on what use is made of
> the information in the ontology. (016)
I agree that Sean's point is important, and I agree with Pat that
more information (i.e., pragmatics and a lot of related background
about engineering practice) is needed before Sean's point can be
suitably defined and quantified. (017)
SB> Unless I can quantify these risk factors, then I will not trust
> the Semantic Web for anything other than low impact actions. (018)
I agree that engineers are expected to quantify the risk factors,
but the information needed to do so depends on many more issues
than just the ontology. (019)
PH> I am confident that the Sweb will survive being un-trusted
> by conservative engineers for a while. (020)
If I were a physician or an engineer who hoped to avoid a malpractice
suit, I would not use *anything* I found on the WWW (semantic or not)
without doing a great deal of "due diligence" on where that data came
from, how it was derived, what are the previous experiences of using
it, etc. (021)
I take my life in my hands whenever I drive my car or fly in a plane,
and I'm willing to continue doing so. But there is no currently
available ontology (not CYC, SUMO, DOLCE, BFO, or anything else I
have seen so far), which I would trust with my own life or anybody
else's (especially if they might have survivors who could sue me). (022)
John (023)
_________________________________________________________________
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 (024)
|