SUO: Re: Graph databases
Jean-Luc,
I agree that a good inference engine is essential. But I would
also add that it must be part of an overall architecture that can
support all three methods of reasoning: induction, deduction,
and abduction. And it must have good methods of assessing the
relevance or plausibility of any of its conclusions.
And note that the inference methods must be clearly distinguished
from the methods for assessing relevance or plausibility. My major
criticism of the fuzzy logic proponents is that they are guilty of
the fallacy of misplaced fuzziness: they confuse the methods of
inference with the methods for assessing relevance. Both are
important, but you can't replace one with the other.
JLD> If you don't have a "state of the art" inference engine you end
> up trying to prove that China isn't a soft drink.
But there are many sad stories making the rounds about so-called
"state of the art" inference engines doing really stupid things.
One of them supposedly got hung up in a loop trying to prove
that Nelson Mandela is not a lawn chair.
That system, by the way, was supposed to have "common sense", but
somehow it still got way off track.
John Sowa