On Wed, April 3, 2013 11:53, John F Sowa wrote: (01)
> ... that is a very common pattern in English:
> [fruit] & [type of plant] => subtype that produces that type of fruit. (02)
In the case of "pineapple tree" or "pine cone", the [type of plant] has
to be generalized. (03)
> The general principle for Noun-Noun compounds is that the speaker has
> some pattern in mind that relates the two nouns and the speaker knows
> (or assumes) that the listener is aware of that pattern and can use
> it to interpret the implicit relationship. (04)
> Examples: 'steamer clam' and 'steamer duck'. (05)
It is best to encode standard phrases such as the above. The derivations
may suggest patterns that are often used, but there is no need to stress
the NL engine by not encoding such phrases that would be defined in
appropriate references. (06)
> ... (07)
> There is no limit to the kinds of pattern. ... (08)
>> Fake diamond is much more complicated to analyze. (09)
> Actually, adjective-noun combinations are usually easier to analyze
> because the kinds of patterns fall into a more conventional range. (010)
> The adjective 'fake' is called *privative* because it deprives
> the thing it modifies of one or more typical attributes. A fake
> diamond has many easily observable attributes of a diamond except
> the essential property of being made of crystallized carbon. (011)
> But note the term 'stuffed bear'. The word 'stuffed' has a privative
> effect of denying the essential property of a bear. But it adds
> some specific information, which the word 'fake' does not. (012)
Here, rules for "stuffed" and "fake" are appropriate. Unless there is
a meaning that is other than that derived from the rules, there would
be no need for encoding individual phrases -- unless a specific phrase
occurs frequently in the domain of an application. (013)
> ... (014)
>> For more fun, consider White tigers, Paper tigers, Paper Airplanes,
>> and Model airplanes. (015)
> 'White tiger' follows the default rules. But 'white elephant' is
> interesting. It follows the same rule as 'white tiger', but it adds
> the background info that white elephants are sacred, but they eat
> a lot and a gift of a white elephant is a liability -- from that
> the term 'white elephant' has become a frozen metaphor for a gift
> that has become a liability. (016)
Standard metaphors should be encoded. It would take a lot of
encoded knowledge to derive the metaphorical sense of "white
elephant". (017)
> Bottom line: To interpret NL modifiers, you need an analogy engine
> that can find and compare widely used patterns. To be efficient,
> you need software that can find analogies in logarithmic time. (018)
To be efficient, standardized phrases should be encoded. The analogy
engine should only be used if a term look-up fails. The analogy engine
should have specific rules for specific patterns based on the classes of
the possible meanings of both the modifying and the modified sub-terms. (019)
-- doug f (020)
> John (021)
_________________________________________________________________
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 (022)
|