Pat, (01)
I certainly did not intend that: (02)
> I was simply reacting, at the beginning of this thread,
> against a naive and uninformed assumption that the Web
> was a direct source of ontological knowledge, or perhaps
> even was a giant ontology all by itself. (03)
I regard it more like the raw material that a lexicographer
would use as a source of citations. Some human being would
have to analyze all the citations for a given word, such as
'deep', in order to write suitable definitions for word senses,
as in 'deep lake', 'deep hole', 'deep voice', or 'deep thought'. (04)
> I have never seen anything to suggest that common sense reasoning
> is 'shallow'. I think that because we all have common sense and
> seem to have acquired it without intellectual effort, we assume
> it must be simple or shallow; whereas in fact it takes about as
> long to acquire all this knowledge (and the concomitant linguistic
> skill that comes along with it) as it does to become a chess master
> or a professional concert violinist or indeed to achieve mastery
> of any other skill: about 10-12 years of constant practice, several
> hours every day. (05)
I agree. I would compare commonsense reasoning to the kind of
case-based reasoning that is done in business, law, and medicine.
Those fields requires many long years of study to acquire the
necessary experience. But the reasoning requires access to an
enormous number of cases, from which the deductive chains are much
shorter than in the typical proofs of modern theorem provers. (06)
> Maybe all this everyday expertise is best used by a case-based
> reasoner which cannot perform logical inferences, but I doubt it. (07)
First of all, a case-based reasoner most definitely does logical
inferences -- although Roger Schank obscured the underlying
principles by forbidding his students to use logic. (08)
One source I cite is Sextus Empiricus (2nd century AD), who made the
following observation: (09)
1. In any empirical subject, any theory used for deduction must have
been obtained by a previous process of induction. (010)
2. Therefore, the accuracy of the deduction is no better than the
adequacy of the set of cases that were used for induction. (011)
Another source is Ibn Taymiyya (14th c AD), who added the following: (012)
3. If case-based reasoning is applied to *exactly* the same cases
that were used for induction, it could generate exactly the same
inferences by a one-step analogy that were obtained by the two
steps of induction + abduction. (013)
I elaborate this discussion in Section 2 of the following paper: (014)
http://www.jfsowa.com/pubs/analog.htm (015)
Summary: (016)
1. In physics, it is possible to derive an enormous number of
consequences from a small number of basic principles. (017)
2. Therefore, theories in physics have proved to be extremely
valuable. (I'll avoid the f----ful word.) (018)
3. But very few subjects one encounters in everyday life (even in
the professions of law and medicine) have theories that remotely
resemble physics. (019)
4. For those subjects, the amount of effort required to derive
theories that cover all the details is enormous, and nobody
does it. I don't believe that it is useful for Lenat and
people with similar inclinations even to attempt it. (020)
5. Instead, I would recommend case-based reasoning, which can be
interpreted as domain-specific induction on just those cases that
are relevant to the immediate problem, followed by short chains
of deduction to adapt the result to the problem at hand. (021)
John (022)
_________________________________________________________________
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 (023)
|