Review: Week 8

Describe Larkin and Simon's model of human cognition. Your description should include terms like short/long term memory, productions.
• In that model, how do we explain expert performance (where experts can do things faster than novices)?
• In that model, how we explain novice performance>?

Use this example for the following questions.

B3:
if	the step is pack-large-items
and	there is a large item to be packed
and	there is a large bottle to be packed
and	there is a bag with < 6 large items
then	put the bottle into the bag

B4:
if	the step is pack-large-items
and	there is a large item to be packed
and	there is a bag with < 6 large items
then	put the large item into the bag

B5:
if	the step is pack-large-items
then	the step is pack-medium-items

1. Define, with an example, Working memory, match, wselectesolve, act
2. What is the point of rule B5?
3. With respect to the BAGGER rules described above, define with examples conflict resolution using:
   • Rule ordering
   • Data ordering
   • Specificity ordering
   • Size ordering (and your answer should explain why size ordering is faster than specificity ordering)

1. "The global and uniform nature of rule-based systems is both a blessing and a curse". Explain, with examples from the history of rule-based systems.
2. Define the RETE network, explaining its theoretical benefits.
3. The two most successful experiments in building and maintaining large rule-based systems were conducted by John McDermott and Paul Compton. These experiments took very difference stances about the nature of the rule maintenance problem:
   • McDermott: if can't use it, don't ask for it.
   • Compton: if it works, don't change it
   Explain these two stances using the following proposed definition of "diagnosis"
4. Compton's ripple down rules take the form

   rule ID1 if condition THEN EXCEPT rule ID2 THEN conclusion because EXAMple
   

   Explain the use of this rule using the following example:
5. Using the above example, explain "difference list" generation.