Lecture 8 - Backward Chaining Flashcards
What is the basic idea of backward chaining rule interpreters?
Start with a conclusion hypothesis
Find rules whose RHS draws conclusions about the hypotheses
Determine if the LHS of these rules is true
If so execute the RHS, confirming or rejecting the hypothesis
With backward chaining rule interpreters, if the LHS of a rule does not match the conclusion, the interpreter will
Set up a subgoal with the LHS as the conclusion and look for rules to match the LHS
Backward chaining is more similar to a human’s way of solving a problem because
It starts with a goal and works its way through the requirements for the goal
“I need X, but to get to X I need Y, which needs me to have solved Z first”
Backward chaining is ______ driven, as opposed to forward chaining which is _______ driven
Goal driven
Data driven
What are the advantages of backward chaining over forward chaining?
May need less working memory, as
Unneeded rules are never executed
Conflict resolution is not necessary as only the rules leading to the conclusion are evaluated