- From: John F. Sowa <sowa@bestweb.net>
- Date: Sat, 05 Sep 2009 02:18:16 -0400
- To: "[ontolog-forum]" <ontolog-forum@ontolog.cim3.net>
- CC: semantic-web@w3.org
Azamat and Rich, AA> Anti-realism to ontological entities, claiming about the > non-reality of nonobservable and nontangible entities with > human senses (currently abstract entities), is becoming > fashionable intellectual style. Scientists tend to be realists about their subject matter. Physicists, for example, usually believe that their laws refer to something that really exists. They assume that unobservable entities such as atoms and fields actually exist. The people who created the difficulties were the 19th century positivists, such as Auguste Comte and Ernst Mach. They claimed that all scientific laws should relate observable measurements and avoid any assumptions or talk about unobservable entities. Mach, for example, believed that thermodynamics should be limited to direct relationships between measurable quantities, such as pressure, volume, and temperature. He fought a life-long battle against Boltzmann's statistical mechanics, because Boltzmann assumed unobservable atoms. With his famous papers of 1905, Einstein destroyed Mach's assumptions: his paper on Brownian motion showed that atoms and molecules could be observed by their effects on tiny particles that could be seen through a microscope. His two papers about relativity and photons assumed unobservable principles and entities that violated all of Mach's rules about how theories should be formulated. Unfortunately, the psychologists didn't have anyone of Einstein's stature to protect them, and the positivists led the behaviorists to change the name of the field to get rid of any talk about a "psyche". That led to half a century of sterile research, popularly known as "rat psychology." RC> A lovely, intuitive description of nominalism that makes > it equivalent to entity tracking in concept space... I'm not sure what you mean by concept space, since concepts are usually defined by intensional methods of definitions and axioms. In any case, data mining is a good example of a nominalist procedure: 1. Data mining starts with a database of low-level facts. 2. It applies well-defined algorithm(s) that analyze the DB to discover patterns in the data. 3. Those patterns might be the result of fundamental laws, or they might be accidental patterns that could be violated by the next update to the database. 4. Some additional analysis and testing is necessary to distinguish principles from coincidences. Points #3 and #4 are critical. Data mining is useful as a method for discovering patterns, but it has no criteria for distinguishing patterns that result from fundamental principles from accidental patterns that result from mere coincidence. That is the weakness of nominalism. John
Received on Saturday, 5 September 2009 06:18:50 UTC