CIS 5603. Artificial Intelligence
Practical Reasoning
1. Knowledge, operation, and goal
"Practical reasoning" refers to the reasoning about "what to do", in contrast with "theoretical reasoning" that is about "what to believe". Their relation corresponds to the relation between declarative knowledge and procedural knowledge.
A system's knowledge/belief relates the system's action/operation and to its goal(s).
2. Logic programming
Logic programming treats problem solving as an inference process.
Example: Prolog (lecture notes).
Related techniques:
3. Planning
Planning: to organize operations to achieve a goal.
Planning can be carried out as state-space search without executing the operations, but recording them.
Planning as reasoning: to specify each state by a set of propositions, and each operation by its precondition and consequences (postconditions). Example: STRIPS. Frame problem: It is hard (if possible) to fully specify an operation.
Planning can be forward, backward, and both. A major issue is to avoid combinatorial explosion.
Scheduling handles the details of timing in plan.
Belief-desire-intention (BDI) model manages plan selection and execution.
Various approaches of self-programming are being explored.
Reading