SUO: Re: Logic, Topic Maps, & RDF

["John F. Sowa" <sowa@bestweb.net>]

Jon,

The semantic distinction you are trying to emphasize
is indeed important.  I have no quarrel with that.

But then we have a separate issue of whether every
semantic distinctions must in every case be reflected
in a syntactic distinction.  I have simply pointed
out that it is common practice -- and I believe
right, fitting, and proper practice -- in all natural
languages not to make explicit distinctions when
those distinctions are obvious.

In English, the difference between the relational
structure of Wife and Woman is obvious because
it can be found in any dictionary (and presumably
in the minds of every English speaker).  In French,
there is no clue whatever in the spelling or the
pronunciation of the word "femme" in the two senses,
but it is present in the dictionary and in feelthy
French postcards.

When we are parsing sentences in English, French,
Russian or any other language, we must design our
software to handle both the syntactic and the
semantic distinctions:

  1. For CGs, we handle the syntactic distinctions
     by keeping the graph structure as close as
     possible to the NL syntax (while keeping the
     graphs syntactically unambiguous).

  2. And we handle the semantic distinctions
     by supporting canonical graphs, which
     make the semantic structure with all the
     relations and implicit participants fully
     explicit.

This approach works very nicely.

 > In that respect, notations that lead
 > us into oblivion are not our friends.

I don't know who you mean by "us".  Our computers
are not led astray.  English, French, and Russian
speakers are not led astray.  I don't see a problem.

John