Re: SUO: Re: Ballot Comment
Adam,
In my previous note, I explained that you can always map the entire
lattice of theories to one giant monolithic theory. Therefore, it
is not possible to "prove" that the giant theory won't work. It is
merely a fact of nature that it is very difficult to keep one big
theory consistent, and extremely difficult for any human being to
comprehend all the possible interactions between axioms of one big
theory.
> Forgive me for speaking somewhat bluntly below. I don't wish to offend
> but I think you really have to be more concrete. You've said a single
> ontology won't work but I think there's a burden of proof on you to show in
> detail why.
Perhaps the word "lattice" makes my proposal sound somehow
"theoretical" or "difficult" or "mathematical".
Let me replace the word "lattice" with "library". What I am proposing
is something that every software engineer is familiar with: a library
of modules that can be imported into the current module upon request.
The major difference is that my proposal also includes three operators
for relating modules: generalization, specialization, and analogy.
It also includes one module at the top (call it the starter module),
which is automatically included in every module, and one module at
the bottom (call it the trash bin), which results from inconsistencies
that cause a theory to degenerate into a mess.
There is nothing difficult or complex about the idea. And those
extra three operators give you the ability to do belief revision,
something which is widely used in the AI community.
John Sowa