SCXT 350

Hour Exam #2

Name_____________________________________

Wednesday, Nov. 6

100 pts.

How do we solve problems? An how do we carry with us useful information about our environment and how to act in it? How could we give a computer program, a robot perhaps, this sort of knowledge representation and memory? And finally, are we moving in the right direction?

I. Problem solving

1. Give definitions of the following terms (5 pts. each):

a. State (as in problem solving)

b. Operator

c. Precondition (as for an operator)

d. Postcondition (as for an operator)

e. Production (as for an expert system)

f. heuristic (as in search)

2. (10 pts.) We discussed problem-solving in blocksworld. States describe the current position of the objects in blocksworld: where blocks are, where the robot arm is, whether or not the robot arm is grasping anything. Give an example of an operator in this environment and say what the preconditions and postconditions for that particular operator are.

4. (15 pts.) Consider the problem space of the wine-pouring problem:

Give a brief description of breadth-first search or depth-first search (but saying which one you are describing), illustrating your response by tracing (giving in order) some of the steps that might be taken in solving the wine-pouring problem above. Feel free to include states not in the portion of the diagram given above.

We carry around with us all the time all sorts of useful information about the world. This information helps us to operate effectively in our environment. How can we store such useful information in a computer?

II. Knowledge Representation

1. (10 pts.) An ISA hierarchy is a simple form of a semantic network. Briefly sketch how the information in the following could be represented in a semantic network:

Birds generally have feathers and fly, except for penguins which are birds but do not fly. In common with other birds, however, penguins do have beaks. Opus is a penguin. While most birds do not carry on conversations, Opus has been known to speak at length.

2. (10 pts.) Long-term knowledge in an expert system is contained in the form of productions, sometimes known as rules. Taking an exam is an expert task. Give (informally, not in CLIPS form) an example of a non-trivial rule that you use in figuring out which problem on an exam to work on next.

3. (5 pts.) One of the advantages that logic has as a knowledge representation system is the fact that it carries along its own inference engine (mathematical proof). Give one disadvantage to logic as a knowledge representation system (we discussed several).

II. Computation and the commitment to Symbolic AI.

a. (5 pts.) One of the commitments of cognitive science (or, at least, the part of it we have been studying this term) is that intelligent action is computable. What role does the Turing-Church hypothesis play in this? (i.e., what is the Turing-Church hypothesis, and why is it important to CogSci?)

b. (10 pts. – be sure to look at the next two questions before answering this one)

State the Physical Symbol System Hypothesis

c. (5 pts.) In your own words, as you would explain this to an intelligent friend, what does "sufficient" in the PSSH mean?

d. (5 pts.) Again in your own words, what does "necessary" mean in the PSSH?