[Top] [All Lists]

Re: [ontolog-forum] Event Ontology

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "John F. Sowa" <sowa@xxxxxxxxxxx>
Date: Sun, 06 Sep 2009 20:19:05 -0400
Message-id: <4AA45179.5050701@xxxxxxxxxxx>
Mike, Ravi, and Rich,    (01)

The primary point I was trying to make is that the distinction
between nominalism and realism depends on what you do *after* you
form the hypothesis.  Data mining (no matter what algorithm is used
or whether the procedure is carried out by computer or by humans)
is the process of analyzing a given set of data to find patterns.    (02)

David Hume correctly noted that there is no way to tell whether a
hypothesis formed by analyzing some data represents real cause and
effect relationships or whether it is a mere coincidence.    (03)

But what Hume did not consider is the predictive value of a law.
Before any hypothesis can be considered a law, it must be tested
by the predictions it makes for new experiments that derive new data
unrelated to the original data from which the hypothesis was derived.
If a hypothesis consistently makes correct predictions, scientists
conclude that it says something true about some aspect of reality.    (04)

That is the critical point that Hume and other nominalists ignored:
the requirement that any proposed law of nature must be able to
make testable predictions under conditions that are different
from the ones that produced the original data was derived.    (05)

MB> Hagar: "What causes disease, Doctor Zook?"
 > Dr. Zook: "Disease is caused by tiny organisms that are invisible
 > to the naked eye"
 > Hagar: ?
 > Dr. Zook: "They're called bad fairies."    (06)

It's almost irrelevant what you call it.  Just making the hypothesis
has the heuristic value of suggesting further experiments -- i.e.,
developing microscopes or trying to cultivate colonies that can be
seen by the naked eye.    (07)

But the crucial point is that you have to do the experiments.  And if
repeated experiments consistently produce the predicted results, that
implies that something about the hypothesis is really there -- although
there may be significant surprises when it is discovered.    (08)

RC> So you consider data mining to be properly mathematized already?    (09)

As I said, the kind of algorithm is irrelevant, whether it's formal
or informal is irrelevant, and whether it's performed by computer
or by humans is irrelevant.  The distinction between nominalism and
realism depends crucially on further predictions from the hypothesis:    (010)

  - Data mining without further predictions and testing is nominalism.    (011)

  - Data mining (or any other method of forming hypotheses) followed
    by predictions that are tested by further experiment is the basis
    for discovering principles that have some degree of reality.    (012)

RC> The stem cell - a hypothesis about the current state.
 > ...
 > Otherwise, the weak law of large numbers wouldn't work.    (013)

I have no idea why you mentioned those points or why you thought
they were relevant to the discussion.    (014)

RC> There is a missing "is" in the above last block of text.  That's
 > an example of what I mean by "category error".    (015)

I call it a typo.    (016)

John    (017)

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
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (018)

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