This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: Massachusetts Institute of Technology 16.410-13 Principles of Autonomy and Decision Making Assignment #1, tutorial. Due: LEC #4 Problem PS.1.1.1: A simple forward-chaining rule set You are given the following database of assertions: (P a b) (P a c) (P c d) (P d e) (P d f) and the following rules: (R1 IF (P ?x ?y) (P ?y ?z) THEN (G ?x ?z)) (R2 IF (G ?x ?y) (P ?y ?z) THEN (H ?x ?z)) Part 1: New rules In the space below, write all the (new) assertions that will be added to the database by a forward-chaining system. Be sure to write the assertions as lists, i.e. in the same format as the input assertions shown above. Part 2: Triggering 1. How many triggered rule instances are found before the first rule is fired? 2. How many of these triggered rule instances are found again after the first rule is fired? Problem PS.1.1.2: IF-THEN Forward Chaining The following problems refer to general properties of forward-chaining systems using only deduction rules, that is, rules that have only IF and THEN clauses. For this problem, assume the rules do not use STOP assertions anywhere. Part 1: Order Given a rule set with only IF-THEN rules, indicate which of the following statements are true. 1. Nothing that an IF-THEN rule adds to the database can make a rule that is triggered once become untriggered. 2. In our default forward-chainer, assuming no infinite looping, the order of the rules will never affect which assertions end up being added to the database. 3. The following rule can cause infinite looping IF ?x THEN ?x 4. The following rule can cause infinite looping IF THEN (Hello) 5. The following rule can cause infinite looping IF ?x THEN (P ?x) Part 2: Conflict Given a rule set with only IF-THEN rules and assuming no infinite looping, which of the following conflict-resolution strategies will produce the same set of new assertions as the strategy implemented in our default forward-chaining system? 1. Fire all triggered rule instances. 2. Fire the triggered rule instance derived from the most recently added assertion (settle ties randomly) 3. Fire the triggered rule instance derived from the least recently added assertion (settle ties randomly) 4. Fire a random triggered rule instance. True/False problems do not have a Check button. Problem PS.1.1.3: Finding the way Impressed with the argument that all intelligent behavior may arise from rule-based processing, you decide to implement a rule-based system that can navigate around Cambridge. Assume that you have a representation of a map in which every intersection is named, e.g. Harvard-square or place-946 , and in which the existence of street segments that connect intersections is represented by assertions of the form (connect-from place-946 Harvard-square) Consider a small version of the Cambridge map, that consists of just the following places and connections: (connect-from tech-square central-square) (connect-from tech-square inman-square) (connect-from tech-square cambridge-port) (connect-from inman-square porter-square) (connect-from inman-square central-square) (connect-from harvard-square fresh-pond)...
View Full Document
This note was uploaded on 11/07/2011 for the course AERO 16.410 taught by Professor Brianwilliams during the Fall '05 term at MIT.
- Fall '05