j***n 发帖数: 301 | 1 In mathematics, one sometimes lives under the illusion that there is just on
e
logic that formalizes the correct principles of mathematical reasoning, the
socalled
predicate calculus or classical rst-order logic. By contrast, in philosophy
and computer science, one nds the opposite: there is a vast array of logics
for reasoning in a variety of domains. We mention intuitionistic logic, sort
ed
logic, modal logic, description logic, temporal logic, belief logic, dynamic
logic,
Hoare logic, specica | t**k 发帖数: 260 | 2 I guess one reason is that, in CS you have to worry about computational
complexity. Although first-order predicate logic subsumes some of the logics
you mentioned, it's semi-decidable. To be practical, we need those less
expressive but more efficient logics, e.g., modal logic and description
logic.
For non-monotonic logic, it was proposed because in AI systems we may often
have to change our previous belief. In mathematics, you don't need to do
this. |
|