SUO: *Date 30 Mar 2002 -- Translations Among Vocabularies
¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤
SUO WG Members,
A couple of heuristics for approaching a difficult problem are these:
1. Try to simplify the problem.
2. Try to generalize the problem.
There is also a kind of mixed strategy, where one looks for a simpler
version of the problem, but then looks at it in a greater generality.
The problem of inter-comminication can be stated this way:
"How do we know when different people are saying the same thing?"
If I think about the problem of inter-communication in this light,
keeping in mind the sorts of problems that we have even trying to
inter-communicate among ourselves about it, then my first bit of
subgoaling would be to look at a version of it that arises even
in a very simple "zeroth order logical" setting, where we just
have a small vocabulary of monadic terms or a "phrase book"
of stock sentences, taken whole without further analysis.
That would be the simplification phase of the attack.
If we leave it there then we are just talking about
propositional calculus, and the question about when
different expressions mean the same thing or not is
really just the ordinary sort of question about when
different expressions are logically equivalent or not.
But now to make it a bit more interesting, and more realistic
from what we have seen, let us suppose that different people
are using different vocabularies of stock terms or phrases.
What does it take to have a well-posed problem here?
That is to say, what extra data must be given for us
to have a chance of solving the inter-communication
problem in this kind of setting. If we don't have
that much then we are probably wasting our time.
I do not know the most general answer here, but a certain analogy with
the brands of problems that we call "solving systems of equations" in
linear algebra leads me to think along the following lines: We could
have logical equations between a term in one vocabulary and a set of
terms in the other vocabulary. At this point two different kinds of
questions arise: The empirical kind of question is how do we derive
these equations by looking at what people say and do; The rational
kind of question is how do we deduce the logical consequence of the
equations that we find to obtain.
Jon Awbrey
¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤