Unit 5 JWFILES PDF

www.StudentRockStars.
com
om
Logical agents
.c
rs
Unit 5
ta
kS
oc
R
nt
de
tu
.S
w
w
w
www.StudentRockStars.com
Outline
♦ Knowledge-based agents
♦ Wumpus world
om
♦ Logic in general—models and entailment
.c
rs
♦ Propositional (Boolean) logic
ta
kS
♦ Equivalence, validity, satisfiability
oc
R
♦ Inference rules and theorem proving
nt
– forward chaining de
– backward chaining
tu
.S
– resolution
w
w
w
Chapter 7 2
Knowledge bases
Inference engine domain−independent algorithms
Knowledge base domain−specific content
om
.c
Knowledge base = set of sentences in a formal language
rs
ta
Declarative approach to building an agent (or other system):
kS
Tell it what it needs to know
oc
R
Then it can Ask itself what to do—answers should follow from the KB
nt
de
Agents can be viewed at the knowledge level
tu
i.e., what they know, regardless of how implemented

.S
w
Or at the implementation level www.StudentRockStars.com

w
w
i.e., data structures in KB and algorithms that manipulate them
A simple knowledge-based agent
function KB-Agent( percept) returns an action

static: KB, a knowledge base
om
t, a counter, initially 0, indicating time
Tell(KB, Make-Percept-Sentence( percept, t))
.c
rs
action ← Ask(KB, Make-Action-Query(t))
ta
Tell(KB, Make-Action-Sentence(action, t))
kS
t←t + 1
oc
return action
R
nt
The agent must be able to: de
tu
Represent states, actions, etc.
.S
Incorporate new percepts

w
Update internal representations of the world

w
w
Deduce hidden properties of the world

Deduce appropriate actions
Chapter 7 4
Wumpus World PEAS description

Performance measure
gold +1000, death -1000
-1 per step, -10 for using the arrow
om
Breeze
Environment 4 Stench
PIT
.c
Squares adjacent to wumpus are smelly
rs
Breeze
Breeze
ta
Squares adjacent to pit are breezy 3 Stench PIT
kS
Gold
Glitter iff gold is in the same square Stench Breeze
oc
2
Shooting kills wumpus if you are facing it
R
Shooting uses up the only arrow
nt
Breeze Breeze
1 PIT
Grabbing picks up gold if in same square
de START
Releasing drops the gold in same square

tu
1 2 3 4
.S
Actuators Left turn, Right turn, www.StudentRockStars.com

w
w
Forward, Grab, Release, Shoot

w
Sensors Breeze, Glitter, Smell
Chapter 7 5
Wumpus world characterization

Observable??
om
.c
rs
ta
kS
oc
R
nt
de
tu
.S
w
w
w
Chapter 7 6

Observable?? No—only local perception
Deterministic??
om
.c
rs
ta
kS
oc
R
nt
de
tu
.S
w
w
w
Chapter 7 7

Deterministic?? Yes—outcomes exactly specified
om
Episodic??
.c
rs
ta
kS
oc
R
nt
de
tu
.S
w
w
w
Chapter 7 8

om
Episodic?? No—sequential at the level of actions
.c
rs
Static??
ta
kS
oc
R
nt
de
tu
.S
w
w
w
Chapter 7 9

om
.c
rs
Static?? Yes—Wumpus and Pits do not move
ta
kS
Discrete??
oc
R
nt
de
tu
.S
w
w
w
Chapter 7 10

om
.c
rs
ta
kS
Discrete?? Yes
oc
R
Single-agent??
nt
de
tu
.S
w
w
w
Chapter 7 11

om
.c
rs
ta
kS
Discrete?? Yes
oc
R
Single-agent?? Yes—Wumpus is essentially a natural feature
nt
de
tu
.S
w
w
w
Chapter 7 12
Exploring a wumpus world
om
.c
rs
ta
kS
oc
R
OK
nt
de
tu
.S
w
OK OK
w
w
Chapter 7 13
om
.c
rs
ta
kS
oc
R
B OK
nt
de
A
tu
.S
w
OK OK
w
w
Chapter 7 14
om
.c
rs
P?
ta
kS
oc
R
B OK P?
nt
de
A
tu
.S
w
OK OK
w
w
Chapter 7 15
om
.c
rs
P?
ta
kS
oc
R
B OK P?
nt
de
A
tu
.S
w
OK S OK
w
w
A A
Chapter 7 16
om
.c
rs
P?
ta
P
kS
oc
R
B OK P?
nt
de OK
A
tu
.S
w
OK S OK
w
W
w
A A
Chapter 7 17
om
.c
rs
P?
ta
P
kS
oc
R
B OK P?
nt
de OK
A A
tu
.S
w
OK S OK
w
W
w
A A
Chapter 7 18
om
.c
rs
P? OK
ta
P
kS
oc
R
B OK P? OK
nt
de OK
A A
tu
.S
w
OK S OK
w
W
w
A A
Chapter 7 19
om
.c
rs
P? OK
ta
P
kS
oc
R
B OK P? BGS OK
nt
de OK
A A A
tu
.S
w
OK S OK
w
W
w
A A
Chapter 7 20
Other tight spots
P?
om
Breeze in (1,2) and (2,1)
.c
B OK P? ⇒ no safe actions
rs
P?
A
ta
Assuming pits uniformly distributed,
kS
OK B OK
(2,2) has pit w/ prob 0.86, vs. 0.31
oc
P?
A A
R
nt
de
Smell in (1,1)
tu
⇒ cannot move
.S
w
Can use a strategy of coercion:

w
shoot straight ahead

w
S wumpus was there ⇒ dead ⇒ safe

A wumpus wasn’t there ⇒ safe
Chapter 7 21
Logic in general
Logics are formal languages for representing information
such that conclusions can be drawn
om
Syntax defines the sentences in the language
.c
rs
Semantics define the “meaning” of sentences;
ta
i.e., define truth of a sentence in a world
kS
oc
E.g., the language of arithmetic
R
nt
x + 2 ≥ y is a sentence; x2 + y > is not a sentence
de
tu
x + 2 ≥ y is true iff the number x + 2 is no less than the number y
.S
w
x + 2 ≥ y is true in a world where x = 7, y = 1

w
x + 2 ≥ y is false in a world where x = 0, y = 6

w
Chapter 7 22
Entailment
Entailment means that one thing follows from another:
KB |= α
om
Knowledge base KB entails sentence α
.c
if and only if
rs
ta
α is true in all worlds where KB is true
kS
oc
E.g., the KB containing “the Giants won” and “the Reds won”
R
entails “Either the Giants won or the Reds won”
nt
de
E.g., x + y = 4 entails 4 = x + y
tu
.S
Entailment is a relationship between sentences (i.e., syntax)

w
that is based on semantics

w
w
Note: brains process syntax (of some sort)
Chapter 7 23
Models
Logicians typically think in terms of models, which are formally
structured worlds with respect to which truth can be evaluated
om
We say m is a model of a sentence α if α is true in m
.c
M (α) is the set of all models of α
rs
ta
kS
Then KB |= α if and only if M (KB) ⊆ M (α)
x
oc
x x x
E.g. KB = Giants won and Reds won x
x x
x
R
x
M( ) x
α = Giants won
nt
x x x
x
x x
de x x
x
x
x
x
x
x x
tu
x x x x
xx x xx
.S
x
x x
x x x
w
M(KB) x
w
x x x
x x
w
Chapter 7 24
Entailment in the wumpus world
om
.c
rs
Situation after detecting nothing in [1,1],
ta
moving right, breeze in [2,1]
? ?
kS
oc
B
R
Consider possible models for ?s ?
nt
A A
assuming only pits

de
tu
.S
3 Boolean choices ⇒ 8 possible models

w
w
w
Chapter 7 25
Wumpus models
2 PIT
2
om
Breeze
1
Breeze
1 PIT
1 2 3
.c
1 2 3
rs
ta
2 PIT
kS
2 PIT
2
Breeze
1 PIT
oc
Breeze
1
Breeze
1
1 2 3
1 2 3
R
1 2 3
nt
2 PIT
de 2 PIT PIT
tu
Breeze
1
Breeze
1 PIT 2 PIT PIT
.S
1 2 3
1 2 3
w
Breeze
1 PIT
w
1 2 3
w
Chapter 7 26
Wumpus models
2 PIT
2
om
Breeze
1
Breeze
1 PIT
1 2 3
.c
1 2 3
KB
rs
ta
2 PIT
kS
2 PIT
2
Breeze
1 PIT
oc
Breeze
1
Breeze
1
1 2 3
1 2 3
R
1 2 3
nt
2 PIT
de 2 PIT PIT
tu
Breeze
1
Breeze
1 PIT 2 PIT PIT
.S
1 2 3
1 2 3
w
Breeze
1 PIT
w
1 2 3
w
KB = wumpus-world rules + observations
Chapter 7 27
Wumpus models
2 PIT
2
om
Breeze
1
Breeze
1 PIT
1 2 3
.c
1 2 3
KB
rs
1
ta
2 PIT
kS
2 PIT
2
Breeze
1 PIT
oc
Breeze
1
Breeze
1
1 2 3
1 2 3
R
1 2 3
nt
2 PIT
de 2 PIT PIT
tu
Breeze
1
Breeze
1 PIT 2 PIT PIT
.S
1 2 3
1 2 3
w
Breeze
1 PIT
w
1 2 3
w

α1 = “[1,2] is safe”, KB |= α1, proved by model checking
Chapter 7 28
Wumpus models
2 PIT
2
om
Breeze
1
Breeze
1 PIT
1 2 3
.c
1 2 3
KB
rs
ta
2 PIT
kS
2 PIT
2
Breeze
1 PIT
oc
Breeze
1
Breeze
1
1 2 3
1 2 3
R
1 2 3
nt
2 PIT
de 2 PIT PIT
tu
Breeze
1
Breeze
1 PIT 2 PIT PIT
.S
1 2 3
1 2 3
w
Breeze
1 PIT
w
1 2 3
w
Chapter 7 29
Wumpus models
2 PIT
2
om
Breeze
1
Breeze
1 PIT
2
1 2 3
.c
1 2 3
KB
rs
ta
2 PIT
2 PIT
kS
2
Breeze
Breeze
1 PIT
oc
1
Breeze
1
1 2 3
1 2 3
R
1 2 3
nt
2 PIT PIT
2 PIT de Breeze
1
tu
Breeze
1 PIT 2 PIT PIT
1 2 3
.S
1 2 3
Breeze
1 PIT
w
1 2 3
w
w

α2 = “[2,2] is safe”, KB 6|= α2
Chapter 7 30
Inference
KB ì α = sentence α can be derived from KB by procedure i
Consequences of KB are a haystack; α is a needle.
om
Entailment = needle in haystack; inference = finding it
.c
Soundness: i is sound if
rs
ta
whenever KB ì α, it is also true that KB |= α
kS
www.StudentRock Stars.com
oc
Completeness: i is complete if
R
whenever KB |= α, it is also true that KB ì α
nt
de
Preview: we will define a logic (first-order logic) which is expressive enough
tu
to say almost anything of interest, and for which there exists a sound and
.S
complete inference procedure.

w
w
w
That is, the procedure will answer any question whose answer follows from
what is known by the KB.
Chapter 7 31
Propositional logic: Syntax

Propositional logic is the simplest logic—illustrates basic ideas
The proposition symbols P1, P2 etc are sentences
om
If S is a sentence, ¬S is a sentence (negation)
.c
rs
If S1 and S2 are sentences, S1 ∧ S2 is a sentence (conjunction)
ta
kS
If S1 and S2 are sentences, S1 ∨ S2 is a sentence (disjunction)
oc
R
If S1 and S2 are sentences, S1 ⇒ S2 is a sentence (implication)
nt
de
If S1 and S2 are sentences, S1 ⇔ S2 is a sentence (biconditional)
tu
.S
w
w
w
Chapter 7 32
Propositional logic: Semantics

Each model specifies true/false for each proposition symbol
E.g. P1,2 P2,2 P3,1
om
true true f alse
.c
(With these symbols, 8 possible models, can be enumerated automatically.)
rs
ta
kS
Rules for evaluating truth with respect to a model m:
oc
¬S is true iff S is false
R
S1 ∧ S2 is true iff is true and is true
nt
S1 S2
S1 ∨ S 2 is true iff
de
S1 is true or S2 is true
tu
S1 ⇒ S 2 is true iff S1 is false or S2 is true
.S
i.e., is false iff S1 is true and S2 is false

w
w
S1 ⇔ S 2 is true iff S1 ⇒ S2 is true and S2 ⇒ S1 is true

w
Simple recursive process evaluates an arbitrary sentence, e.g.,

¬P1,2 ∧ (P2,2 ∨ P3,1) = true ∧ (f alse ∨ true) = true ∧ true = true
Chapter 7 33
Truth tables for connectives

P Q ¬P P ∧Q P ∨Q P ⇒Q P ⇔Q
false false true false false true true
false true true false true true false
om
true false false false true false false
.c
true true false true true true true
rs
ta
kS
oc
R
nt
de
tu
.S
w
w
w
Chapter 7 34
Wumpus world sentences

Let Pi,j be true if there is a pit in [i, j].
Let Bi,j be true if there is a breeze in [i, j].
om
¬P1,1
.c
¬B1,1
rs
B2,1
ta
kS
“Pits cause breezes in adjacent squares”
oc
R
nt
de
tu
.S
w
w
w
Chapter 7 35
Wumpus world sentences

Let Pi,j be true if there is a pit in [i, j].
Let Bi,j be true if there is a breeze in [i, j].
om
¬P1,1
.c
¬B1,1
rs
B2,1
ta
kS
“Pits cause breezes in adjacent squares”
oc
R
B1,1 ⇔ (P1,2 ∨ P2,1)
nt
B2,1 ⇔
de
(P1,1 ∨ P2,2 ∨ P3,1)
tu
.S
“A square is breezy if and only if there is an adjacent pit”

w
w
w
Chapter 7 36
Truth tables for inference

B1,1 B2,1 P1,1 P1,2 P2,1 P2,2 P3,1 R1 R2 R3 R4 R5 KB
false false false false false false false true true true true false false
false false false false false false true true true false true false false
om
... ... ... ... ... ... ... ... ... ... ... ... ...
.c
false true false false false false false true true false true true false
rs
false true false false false false true true true true true true true
ta
false true false false false true false true true true true true true
kS
false true false false false true true true true true true true true
oc
false true false false true false false true false false true true false
R
... ... ... ... ... ... ... ... ... ... ... ... ...
nt
true true true true true true
de true false true true false true false
tu
Enumerate rows (different assignments to symbols),
.S
if KB is true in row, check that α is too www.StudentRockStars.com

w
w
w
Chapter 7 37
Inference by enumeration
Depth-first enumeration of all models is sound and complete
function TT-Entails?(KB, α) returns true or false
om
inputs: KB, the knowledge base, a sentence in propositional logic
.c
α, the query, a sentence in propositional logic
rs
symbols ← a list of the proposition symbols in KB and α
ta
return TT-Check-All(KB, α, symbols, [ ])
kS
oc
function TT-Check-All(KB, α, symbols, model) returns true or false
R
if Empty?(symbols) then
nt
de
if PL-True?(KB, model) then return PL-True?(α, model)
tu
else return true
.S
else do
w
P ← First(symbols); rest ← Rest(symbols)

w
return TT-Check-All(KB, α, rest, Extend(P , true, model)) and

w
TT-Check-All(KB, α, rest, Extend(P , false, model))
O(2n) for n symbols; problem is co-NP-complete

Chapter 7 38
Logical equivalence
Two sentences are logically equivalent iff true in same models:
α ≡ β if and only if α |= β and β |= α
om
(α ∧ β) ≡ (β ∧ α) commutativity of ∧
.c
(α ∨ β) ≡ (β ∨ α) commutativity of ∨
rs
((α ∧ β) ∧ γ) ≡ (α ∧ (β ∧ γ)) associativity of ∧
ta
((α ∨ β) ∨ γ) ≡ (α ∨ (β ∨ γ)) associativity of ∨
kS
¬(¬α) ≡ α double-negation elimination
oc
R
(α ⇒ β) ≡ (¬β ⇒ ¬α) contraposition
nt
(α ⇒ β) ≡ (¬α ∨ β) implication elimination
de
(α ⇔ β) ≡ ((α ⇒ β) ∧ (β ⇒ α)) biconditional elimination
tu
¬(α ∧ β) ≡ (¬α ∨ ¬β) De Morgan

.S
w
¬(α ∨ β) ≡ (¬α ∧ ¬β) De Morgan

w
(α ∧ (β ∨ γ)) ≡ ((α ∧ β) ∨ (α ∧ γ)) distributivity of ∧ over ∨

w
(α ∨ (β ∧ γ)) ≡ ((α ∨ β) ∧ (α ∨ γ)) distributivity of ∨ over ∧
Chapter 7 39
Validity and satisfiability

A sentence is valid if it is true in all models,
e.g., T rue, A ∨ ¬A, A ⇒ A, (A ∧ (A ⇒ B)) ⇒ B
om
Validity is connected to inference via the Deduction Theorem:
.c
KB |= α if and only if (KB ⇒ α) is valid
rs
ta
A sentence is satisfiable if it is true in some model
kS
e.g., A ∨ B, C
oc
R
A sentence is unsatisfiable if it is true in no models
nt
e.g., A ∧ ¬A de
tu
Satisfiability is connected to inference via the following:
.S
KB |= α if and only if (KB ∧ ¬α) is unsatisfiable

w
w
i.e., prove α by reductio ad absurdum

w
Chapter 7 40
Proof methods
Proof methods divide into (roughly) two kinds:
om
Application of inference rules
.c
– Legitimate (sound) generation of new sentences from old
rs
– Proof = a sequence of inference rule applications
ta
Can use inference rules as operators in a standard search alg.
kS
– Typically require translation of sentences into a normal form
oc
R
Model checking
nt
de
truth table enumeration (always exponential in n)
tu
improved backtracking, e.g., Davis–Putnam–Logemann–Loveland
.S
heuristic search in model space (sound but incomplete)

w
e.g., min-conflicts-like hill-climbing algorithms

w
w
Chapter 7 41
Forward and backward chaining

Horn Form (restricted)
KB = conjunction of Horn clauses
Horn clause =
om
♦ proposition symbol; or
.c
♦ (conjunction of symbols) ⇒ symbol
rs
ta
E.g., C ∧ (B ⇒ A) ∧ (C ∧ D ⇒ B)
kS
Modus Ponens (for Horn Form): complete for Horn KBs
oc
R
α1 , . . . , α n , α1 ∧ · · · ∧ α n ⇒ β
nt
β
de
tu
Can be used with forward chaining or backward chaining.

.S
These algorithms are very natural and run in linear time

w
w
w
Chapter 7 42
Forward chaining
Idea: fire any rule whose premises are satisfied in the KB,
add its conclusion to the KB, until query is found
om
Q
P ⇒ Q
.c
rs
L∧M ⇒ P
ta
P
kS
B∧L ⇒ M
oc
A∧P ⇒ L M
R
A∧B ⇒ L
nt
de L
A
tu
B
.S
A B
w
w
w
Chapter 7 43
Forward chaining algorithm
function PL-FC-Entails?(KB, q) returns true or false

inputs: KB, the knowledge base, a set of propositional Horn clauses
om
q, the query, a proposition symbol
local variables: count, a table, indexed by clause, initially the number of premises
.c
inferred, a table, indexed by symbol, each entry initially false
rs
agenda, a list of symbols, initially the symbols known in KB
ta
kS
while agenda is not empty do
oc
p ← Pop(agenda)
R
unless inferred[p] do
nt
inferred[p] ← true de
for each Horn clause c in whose premise p appears do
tu
decrement count[c]
.S
if count[c] = 0 then do
w
if Head[c] = q then return true

w
w
Push(Head[c], agenda)
return false
Chapter 7 44
Forward chaining example
om
.c
P
rs
ta
2
kS
oc
M
R
nt
de 2
L
tu
.S
w
2 2
w
w
A B
Chapter 7 45
om
.c
P
rs
ta
2
kS
oc
M
R
nt
de 2
L
tu
.S
w
1 1
w
w
A B
Chapter 7 46
om
.c
P
rs
ta
2
kS
oc
M
R
nt
de 1
L
tu
.S
w
1 0
w
w
A B
Chapter 7 47
om
.c
P
rs
ta
1
kS
oc
M
R
nt
de 0
L
tu
.S
w
1 0
w
w
A B
Chapter 7 48
om
.c
P
rs
ta
0
kS
oc
M
R
nt
de 0
L
tu
.S
w
1 0
w
w
A B
Chapter 7 49
om
0
.c
P
rs
ta
0
kS
oc
M
R
nt
de 0
L
tu
.S
w
0 0
w
w
A B
Chapter 7 50
om
0
.c
P
rs
ta
0
kS
oc
M
R
nt
de 0
L
tu
.S
w
0 0
w
w
A B
Chapter 7 51
om
.c
P
rs
ta
0
kS
oc
M
R
nt
de 0
L
tu
.S
w
0 0
w
w
A B
Chapter 7 52
Proof of completeness
FC derives every atomic sentence that is entailed by KB
1. FC reaches a fixed point where no new atomic sentences are derived
om
2. Consider the final state as a model m, assigning true/false to symbols
.c
rs
3. Every clause in the original KB is true in m
ta
kS
Proof: Suppose a clause a1 ∧ . . . ∧ ak ⇒ b is false in m
oc
Then a1 ∧ . . . ∧ ak is true in m and b is false in m
R
Therefore the algorithm has not reached a fixed point!
nt
de
4. Hence m is a model of KB
tu
.S
5. If KB |= q, q is true in every model of KB, including m

w
w
General idea: construct any model of KB by sound inference, check α

w
www.StudentRockStars.c
om
Chapter 7 53
Backward chaining
Idea: work backwards from the query q:
to prove q by BC,
check if q is known already, or
om
prove by BC all premises of some rule concluding q
.c
rs
Avoid loops: check if new subgoal is already on the goal stack
ta
kS
Avoid repeated work: check if new subgoal
oc
1) has already been proved true, or
R
2) has already failed
nt
de
tu
.S
w
w
w
Chapter 7 54
Backward chaining example
om
.c
P
rs
ta
kS
oc
M
R
nt
de
L
tu
.S
w
w
w
A B
Chapter 7 55
om
.c
P
rs
ta
kS
oc
M
R
nt
de
L
tu
.S
w
w
w
A B
Chapter 7 56
om
.c
P
rs
ta
kS
oc
M
R
nt
de
L
tu
.S
w
w
w
A B
Chapter 7 57
om
.c
P
rs
ta
kS
oc
M
R
nt
de
L
tu
.S
w
w
w
A B
Chapter 7 58
om
.c
P
rs
ta
kS
oc
M
R
nt
de
L
tu
.S
w
w
w
A B
Chapter 7 59
om
.c
P
rs
ta
kS
oc
M
R
nt
de
L
tu
.S
w
w
w
A B
Chapter 7 60
om
.c
P
rs
ta
kS
oc
M
R
nt
de
L
tu
.S
w
w
w
A B
Chapter 7 61
om
.c
P
rs
ta
kS
oc
M
R
nt
de
L
tu
.S
w
w
w
A B
Chapter 7 62
om
.c
P
rs
ta
kS
oc
M
R
nt
de
L
tu
.S
w
w
w
A B
Chapter 7 63
om
.c
P
rs
ta
kS
oc
M
R
nt
de
L
tu
.S
w
w
w
A B
Chapter 7 64
om
.c
P
rs
ta
kS
oc
M
R
nt
de
L
tu
.S
w
w
w
A B
Chapter 7 65
Forward vs. backward chaining

FC is data-driven, cf. automatic, unconscious processing,
e.g., object recognition, routine decisions
om
May do lots of work that is irrelevant to the goal
.c
rs
BC is goal-driven, appropriate for problem-solving,
ta
e.g., Where are my keys? How do I get into a PhD program?
kS
oc
Complexity of BC can be much less than linear in size of KB
R
nt
de
tu
.S
w
w
w
Chapter 7 66
Resolution
Conjunctive Normal Form (CNF—universal)
conjunction of disjunctions
| {z
of literals}
clauses
om
E.g., (A ∨ ¬B) ∧ (B ∨ ¬C ∨ ¬D)
.c
rs
Resolution inference rule (for CNF): complete for propositional logic
ta
kS
`1 ∨ · · · ∨ `k , m1 ∨ · · · ∨ m n
oc
`1 ∨ · · · ∨ ì−1 ∨ ì+1 ∨ · · · ∨ `k ∨ m1 ∨ · · · ∨ mj−1 ∨ mj+1 ∨ · · · ∨ mn
R
nt
where ì and mj are complementary literals. E.g.,
de
P?
tu
P1,3 ∨ P2,2, ¬P2,2 P
.S
B OK P?
P1,3 OK
w
A A
w
OK S OK
Resolution is sound and complete for propositional logic A A

W
w
Chapter 7 67
Conversion to CNF
B1,1 ⇔ (P1,2 ∨ P2,1)
1. Eliminate ⇔, replacing α ⇔ β with (α ⇒ β) ∧ (β ⇒ α).
om
(B1,1 ⇒ (P1,2 ∨ P2,1)) ∧ ((P1,2 ∨ P2,1) ⇒ B1,1)
.c
rs
2. Eliminate ⇒, replacing α ⇒ β with ¬α ∨ β.
ta
kS
(¬B1,1 ∨ P1,2 ∨ P2,1) ∧ (¬(P1,2 ∨ P2,1) ∨ B1,1)
oc
R
3. Move ¬ inwards using de Morgan’s rules and double-negation:
nt
de
(¬B1,1 ∨ P1,2 ∨ P2,1) ∧ ((¬P1,2 ∧ ¬P2,1) ∨ B1,1)
tu
.S
4. Apply distributivity law (∨ over ∧) and flatten:

w
w
(¬B1,1 ∨ P1,2 ∨ P2,1) ∧ (¬P1,2 ∨ B1,1) ∧ (¬P2,1 ∨ B1,1)

w
Chapter 7 68
Resolution algorithm
Proof by contradiction, i.e., show KB ∧ ¬α unsatisfiable
function PL-Resolution(KB, α) returns true or false
om
inputs: KB, the knowledge base, a sentence in propositional logic
.c
α, the query, a sentence in propositional logic
rs
clauses ← the set of clauses in the CNF representation of KB ∧ ¬α
ta
kS
new ← { }
loop do
oc
for each Ci, Cj in clauses do
R
nt
resolvents ← PL-Resolve(Ci, Cj )
de
if resolvents contains the empty clause then return true
tu
new ← new ∪ resolvents
.S
if new ⊆ clauses then return false

w
clauses ← clauses ∪ new

w
w
Chapter 7 69
Resolution example
KB = (B1,1 ⇔ (P1,2 ∨ P2,1)) ∧ ¬B1,1 α = ¬P1,2
om
P2,1 B1,1 B1,1 P1,2 P2,1 P1,2 B1,1 B1,1 P1,2
.c
rs
ta
B1,1 P1,2 B1,1 P P B1,1 P2,1 B1,1 P P
kS
1,2 2,1 P1,2 1,2 2,1 P2,1 P2,1 P1,2
oc
R
nt
de
tu
.S
w
w
w
Chapter 7 70
Summary
Logical agents apply inference to a knowledge base
to derive new information and make decisions
om
Basic concepts of logic: www.StudentRockStars.com
.c
– syntax: formal structure of sentences
rs
– semantics: truth of sentences wrt models
ta
– entailment: necessary truth of one sentence given another
kS
– inference: deriving sentences from other sentences
oc
– soundess: derivations produce only entailed sentences
R
nt
– completeness: derivations can produce all entailed sentences
de
tu
Wumpus world requires the ability to represent partial and negated informa-
.S
tion, reason by cases, etc.

w
w
Forward, backward chaining are linear-time, complete for Horn clauses

w
Resolution is complete for propositional logic

Propositional logic lacks expressive power
Chapter 7 71

Unit 5 JWFILES PDF

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Unit 5 JWFILES PDF

Uploaded by

Copyright:

Available Formats

www.StudentRockStars.

Knowledge base domain−specific content

i.e., what they know, regardless of how implemented

Or at the implementation level www.StudentRockStars.com

i.e., data structures in KB and algorithms that manipulate them

A simple knowledge-based agent

function KB-Agent( percept) returns an action

Incorporate new percepts

Update internal representations of the world

Deduce hidden properties of the world

Wumpus World PEAS description

Glitter iff gold is in the same square Stench Breeze

Releasing drops the gold in same square

Actuators Left turn, Right turn, www.StudentRockStars.com

Forward, Grab, Release, Shoot

Sensors Breeze, Glitter, Smell

Wumpus world characterization

Wumpus world characterization

Wumpus world characterization

Wumpus world characterization

Wumpus world characterization

Wumpus world characterization

Wumpus world characterization

Exploring a wumpus world

Exploring a wumpus world

Exploring a wumpus world

Exploring a wumpus world

Exploring a wumpus world

Exploring a wumpus world

Exploring a wumpus world

Exploring a wumpus world

Other tight spots

Can use a strategy of coercion:

shoot straight ahead

S wumpus was there ⇒ dead ⇒ safe

x + 2 ≥ y is true in a world where x = 7, y = 1

x + 2 ≥ y is false in a world where x = 0, y = 6

Entailment is a relationship between sentences (i.e., syntax)

that is based on semantics

Note: brains process syntax (of some sort)

Entailment in the wumpus world

assuming only pits

3 Boolean choices ⇒ 8 possible models

KB = wumpus-world rules + observations

KB = wumpus-world rules + observations

KB = wumpus-world rules + observations

KB = wumpus-world rules + observations

complete inference procedure.

Propositional logic: Syntax

Propositional logic: Semantics

i.e., is false iff S1 is true and S2 is false

S1 ⇔ S 2 is true iff S1 ⇒ S2 is true and S2 ⇒ S1 is true

Simple recursive process evaluates an arbitrary sentence, e.g.,

Truth tables for connectives

Wumpus world sentences

Wumpus world sentences

“A square is breezy if and only if there is an adjacent pit”

Truth tables for inference

if KB is true in row, check that α is too www.StudentRockStars.com

function TT-Entails?(KB, α) returns true or false

P ← First(symbols); rest ← Rest(symbols)

return TT-Check-All(KB, α, rest, Extend(P , true, model)) and