Re: Universal Translators

James Rogers (jamesr@best.com)
Mon, 11 Nov 1996 15:00:02 -0800


At 01:07 PM 11/11/96 -0800, you wrote:
>>My reasoning for having a hierarchical (or layered) structure like this
>>is that a single document could then be "read" at varying levels of
>>detail and still be meaningful.

Very interesting. In a sense the language would be somewhat fractal in
nature. I wonder how you would encode something like this? I suppose a
simple version could be done using something like hypertext. You click on a
concept to resolve a particular concept in finer detail, or maybe have a
more global resolution mapping. The plus of this type of heirarchical
language structure is that indexing could be very precise, much more precise
than current single-level linear documents.

Actually, (I am writing this as I am thinking) it might be very nicely
encoded in a type of object-oriented paradigm, similar to object-oriented
programming tools. You could generate (and encode) new concepts and words
as extended instances of a base class. This would force structure upon the
language without really putting any constraints on it. It would probably
also make language processing and comprehension software much easier to
write and model, since it would eliminate ambiguities.

Using an object-oriented paradigm, you could control resolution by selecting
which level of object (concept) inheritance you choose to view. A person
who doesn't want to know the most intimate details of a concept could choose
to view the more generalized base class of that object, and so on. You
could, theoretically, encode the entirety of human knowledge by subclassing
a single base class a very large number of levels. This type of language
encoding would be well suited for computer implementation.

I would be willing to discuss this in great detail with an eye toward
implementation, now that my interest and intellect has been piqued.

-James Rogers
jamesr@best.com