Components of Propositional Logic Artificial Intelligence

Logic constants: True (1), and False (0)

Propositional variables: X = {p1 , q2 , ...}, a set of
Boolean variables
Logical connectives: F = {¬, ∧, ∨, ⇒, ⇔, ...}
Logic and Inferences
Logical sentences: L(X, F ), expressions built from X
and F
Logical interpretations: a mapping θ from X to {0, 1} Readings: Chapter 7 of Russell &
Logical evaluations: a process of applying a mapping θ
to sentences in L(X, F ) (to obtain a value in {0, 1}) Norvig.

Artificial Intelligence – p.2/34 Artificial Intelligence – p.1/34

Properties of Logical Connectives Propositional Variables

∧ and ∨ are commutative Also called Boolean variables, 0-1 variables.
ϕ1 ∧ ϕ2 ≡ ϕ2 ∧ ϕ1 Every statement can be represented by a propositional
ϕ1 ∨ ϕ2 ≡ ϕ2 ∨ ϕ1 variable:
∧ and ∨ are associative p1 = “It is sunny today”
ϕ1 ∧ (ϕ2 ∧ ϕ3 ) ≡ (ϕ1 ∧ ϕ2 ) ∧ ϕ3 p2 = “Tom went to school yesterday”
ϕ1 ∨ (ϕ2 ∨ ϕ3 ) ≡ (ϕ1 ∨ ϕ2 ) ∨ ϕ3 p3 = “f (x) = 0 has two solutions”
∧ and ∨ are mutually distributive p4 = “point A is on line BC ”
ϕ1 ∧ (ϕ2 ∨ ϕ3 ) ≡ (ϕ1 ∧ ϕ2 ) ∨ (ϕ1 ∧ ϕ3 ) p5 = “place a queen at position (1, 2) on a 8 by 8
ϕ1 ∨ (ϕ2 ∧ ϕ3 ) ≡ (ϕ1 ∨ ϕ2 ) ∧ (ϕ1 ∨ ϕ3 ) chessboard”
...
∧ and ∨ are related by ¬ (DeMorgan’s Laws)
¬(ϕ1 ∧ ϕ2 ) ≡ ¬ϕ1 ∨ ¬ϕ2
¬(ϕ1 ∨ ϕ2 ) ≡ ¬ϕ1 ∧ ¬ϕ2
Artificial Intelligence – p.4/34 Artificial Intelligence – p.3/34
 Negation Normal Form (NNF) Properties of Logical Connectives
A propositional sentence is said to be in Negation Normal
Form if it contains no connectives other than ∨, ∧, ¬, and
∧, ⇒, and ⇔ are actually redundant:
if ¬(α) appears in the sentence, then α must be a
propositional variable. ϕ1 ∧ ϕ2 ≡ ¬(¬ϕ1 ∨ ¬ϕ2 )
¬p, p ∨ q, p ∨ (q ∧ ¬r) are NNF. ϕ1 ⇒ ϕ2 ≡ ¬ϕ1 ∨ ϕ2
p ⇔ q, ¬(p ∨ q) are not NNF. ϕ1 ⇔ ϕ2 ≡ (ϕ1 ⇒ ϕ2 ) ∧ (ϕ2 ⇒ ϕ1 )

Every propositional sentence can be transformed into an We keep them all mainly for convenience.
equivalent NNF.
Exercise Use the truth tables to prove all the logical
ϕ1 ⇔ ϕ2 ≡ (ϕ1 ⇒ ϕ2 ) ∧ (ϕ2 ⇒ ϕ1 )
equivalences seen so far.
ϕ1 ⇒ ϕ2 ≡ ¬ϕ1 ∨ ϕ2
¬(ϕ1 ∧ ϕ2 ) ≡ ¬ϕ1 ∨ ¬ϕ2
¬(ϕ1 ∨ ϕ2 ) ≡ ¬ϕ1 ∧ ¬ϕ2 Artificial Intelligence – p.6/34 Artificial Intelligence – p.5/34

Existence of CNF Conjunctive Normal Form (CNF)

A propositional sentence φ is said to be in Conjunctive
Normal Form if φ is True, False, or a conjunction of αi ’s:
Every propositional sentence can be transformed into an
equivalent CNF. That is, from NNF, using α1 ∧ α2 ∧ · · · ∧ αn ,

(ϕ1 ∧ ϕ2 ) ∨ ϕ3 ≡ (ϕ1 ∨ ϕ3 ) ∧ (ϕ2 ∨ ϕ3 ) where each α is a clause (a disjunction of βj ’s):

α = (β1 ∨ β2 ∨ · · · ∨ βm ),

where each β is called a literal, which is either a

variable or the negation of a variable: β = p or β = ¬(p).
¬p, p ∨ q , p ∧ (q ∨ ¬r) are CNF.
p ∨ (q ∧ ¬r) are not CNF.
Every CNF is an NNF but the opposite does not hold.

Artificial Intelligence – p.8/34 Artificial Intelligence – p.7/34

 Interpretations and Models Example: The 8-Queen Problem
A (partial) interpretation is a mapping from variables to {0, 1}. 1 ≤ i, j ≤ 8. qij is true iff a queen is placed in the
Variables: qi,j ,

An complete interpretation is if it maps every variables to {0, 1}. square at row i and column j.
Clauses:
An interpretation θ can be extended to be a mapping from
propositional sentences to {0, 1} if it obeys the following rules: Every row has a queen: For each 1 ≤ i ≤ 8,

θ(¬p) = 1 if and only if θ(p) = 0; qi,1 ∨ qi,2 ∨ qi,3 ∨ qi,4 ∨ qi,5 ∨ qi,6 ∨ qi,7 ∨ qi,8
if θ(p) = 1 or θ(q) = 1 then θ(p ∨ q) = 1;
if θ(p) = 1 and θ(q) = 1 then θ(p ∧ q) = 1; ... Two queens from the same row cannot see each other:

Computing this interpretation is called evaluation. ¬qi,1 ∨ ¬qi,2 , ...

For any propositional sentence Φ, if there exists an interpreation θ
such that θ(Φ) = 1, then θ is a model of Φ and Φ is said to be Two queens from the same column cannot see each other:
satisfiable.
¬q1,j ∨ ¬q2,j , ...
If every complete interpretation is a model for Φ, then Φ is said to be
valid. Two queens on the same diagonal cannot see each other:
Artificial Intelligence – p.10/34 Artificial Intelligence – p.9/34
¬q1,1 ∨ ¬q2,2 , ...

Inference Systems Validity vs. Satisfiability

A sentence is
In practice, an inference system I for PL is a procedure satisfiable if it is true in some interpretation,
that given a set Γ = {α1 , . . . , αm } of sentences and a
valid if it is true in every possible interpretation.
sentence ϕ, may reply “yes”, “no”, or run forever.
If I replies “yes” on input (Γ, ϕ), we say that Γ derives ϕ
in I , Or, I derives ϕ from Γ, or, ϕ is deduced (derived) Reasoning Tools for Propositional Logic:
from Γ in I . and write
A tool to prove a sentence is valid needs only to return
Γ ⊢I ϕ yes or no.

Intuitively, I should be such that it replies “yes” on input A tool to prove a sentence is satisfiable often returns an
(Γ, ϕ) only if ϕ is in fact deduced from Γ by I . interpretation under which the sentence is true.
A sentence φ is valid if and only if ¬φ is unsatisfiable.

Artificial Intelligence – p.12/34 Artificial Intelligence – p.11/34

 All These Fancy Symbols! All These Fancy Symbols!
A∧B ⇒C
is a sentence (an expression built from variables and
In other words,
logical connectives) where ⇒ is a logical connective.
⇒ is a formal symbol of the logic, which is used by the A ∧ B |= C
inference system. is a mathematical abbreviation standing for the statement:
|= is a shorthand for the entailment of formal sentences “every interpretation that makes A ∧ B true, makes C also
that we use to talk about the meaning of formal true.” We say that the sentence A ∧ B entails C.
sentences. A ∧ B ⊢I C
⊢I is a shorthand for the inference procedure that we use is a mathematical abbreviation standing for the statement:
to talk about the output of the inference system I. “I returns yes on input (A ∧ B, C)” [C is deduced from
A ∧ B in I].
The formal symbol ⇒ and the shorthands |=, ⊢I are related.
Artificial Intelligence – p.14/34 Artificial Intelligence – p.13/34

Soundness and Completeness All These Fancy Symbols!

The sentence ϕ1 ⇒ ϕ2 is valid (always true) if and only
if ϕ1 |= ϕ2 .
An inference system I is sound if it derives only
sentences that logically follow from a given set of Example: A ⇒ (A ∨ B) is valid and A |= (A ∨ B)
sentences:
A B A ∨ B A ⇒ (A ∨ B)
if Γ ⊢I ϕ then Γ |= ϕ. 1. False False False True
2. False True True True
An inference system I is complete if it derives all
sentences that logically follow from a given set of 3. True False True True
sentences: 4. True True True True
if Γ |= ϕ then Γ ⊢I ϕ.
or equivalently,
if Γ 6⊢I ϕ then Γ 6|= ϕ.
Artificial Intelligence – p.16/34 Artificial Intelligence – p.15/34
 Inference by Truth Tables Inference in Propositional Logic
The truth-tables-based inference system is sound:
α1 , . . . , αm ⊢T T ϕ implies truth table of (α1 ∧ · · · ∧ αm ) ⇒ ϕ all true
There are two (equivalent) types of inference systems of
implies (α1 ∧ · · · ∧ αm ) ⇒ ϕ is valid
Propositional Logic:
implies (α1 ∧ · · · ∧ αm ) |= ϕ
one based on truth tables (T T )
implies α1 , . . . , αm |= ϕ
one based on derivation rules (R)
It is also complete (exercise: prove it).
Its time complexity is O(2n ) where n is the number of propositional
Truth Tables The inference system T T is specified as
variables in α1 , . . . , αm , ϕ.
follows:
We cannot hope to do better because a related, simpler problem
(determining the satisfiability of a sentence) is NP-complete. {α1 , . . . , αm } ⊢T T ϕ iff all the values in the truth table of
However, the really hard cases of propositional inference when we (α1 ∧ · · · ∧ αm ) ⇒ ϕ are True.
need O(2n ) time are somewhat rare.

Artificial Intelligence – p.18/34 Artificial Intelligence – p.17/34

Rule-Based Inference System Rule-Based Inference System

Inference Rules

And-Introduction An inference system in Propositional Logic can also be

specified as a set R of inference (or derivation) rules.
α β
α∧β Each rule is actually a pattern premises/conclusion.
A rule applies to Γ and derives ϕ if
And-Elimination
α∧β α∧β some of the sentences in Γ match with the premises
of the rule and
α β
ϕ matches with the conclusion.
Or-Introduction
A rule is sound it the set of its premises entails its
α α
conclusion.
α∨β β∨α

Artificial Intelligence – p.20/34 Artificial Intelligence – p.19/34

 Rule-Based Inference Systems Rule-Based Inference System
Inference Rules (cont’d.) Inference Rules (cont’)

Double-Negation-Elimination Implication-Elimination (aka Modus Ponens)

¬¬α α⇒β α
α β
False-Introduction Unit Resolution
α ∧ ¬α α∨β ¬β
False α
False-Elimination Resolution
α∨β ¬β ∨ γ
False α∨γ
β
¬α ⇒ β, β⇒γ
or, equivalently,
¬α ⇒ γ
Artificial Intelligence – p.22/34 Artificial Intelligence – p.21/34

Inference by Proof Inference by Proof

We can represent a proof more visually as a proof tree: We say there is a proof of ϕ from Γ in R if we can derive ϕ
by applying the rules of R repeatedly to Γ and its derived
Example: sentences.
(P ∨ H) ∧ ¬H (P ∨ H) ∧ ¬H
Example: a proof of P from {(P ∨ H) ∧ ¬H}
P ∨H ¬H
1. (P ∨ H) ∧ ¬H by assumption
P
2. P ∨H by ∧-elimination applied to (1)
3. ¬H by ∧-elimination applied to (1)
4. P by unit resolution applied to (2),(3)

Artificial Intelligence – p.24/34 Artificial Intelligence – p.23/34

Soundness of Rule-Based Inferences Rule-Based Inference System
R is sound if all of its rules are sound.
α ∨ β, ¬β ∨ γ
Example: the Resolution rule More formally, there is a proof of ϕ from Γ in R if
α∨γ
1. ϕ ∈ Γ or,
α β γ ¬β α∨β ¬β ∨ γ α∨γ
1. False False False True False True False
2. there is a rule in R that applies to Γ and produces ϕ or,
2. False False True True False True True 3. there is a proof of each ϕ1 , . . . , ϕm from Γ in R and
3. False True False False True False False a rule that applies to {ϕ1 , . . . , ϕm } and produces ϕ.
4. False True True False True True True
5. True False False True True True True
Then, the inference system R is specified as follows:
6. True False True True True True True
7. True True False False True False True Γ ⊢R ϕ iff there is a proof of ϕ from Γ in R
8. True True True False True True True

All the interpretations that make both α ∨ β and ¬β ∨ γ true (ie, 4,5,6,8)
make α ∨ γ also true.
Artificial Intelligence – p.26/34 Artificial Intelligence – p.25/34

One Rule Suffices! The Rules of R

α β α α
Assumptions:
α∧β α∨β β∨α
β ¬β α∨β ¬β
is a special case of when α
False α α∧β α∧β
is False.
α β
α∨β ¬β α∨β ¬β ∨ γ
is a special case of
α α∨γ
α⇒β α α∨β ¬β α∨β ¬β ∨ γ
when γ is False.
β α α∨γ
To prove that Γ ⇒ ϕ is valid,
Convert Γ ∧ ¬(ϕ) into a set of clauses. ¬¬α α ∧ ¬α False
Repeatedly apply the Resolution Rule on the clauses. α False β
If False is derived out, then (Γ ⇒ ϕ) is valid.

Artificial Intelligence – p.28/34 Artificial Intelligence – p.27/34

 Proof by Contradiction Resolution Proof

Γ ∧ ¬ϕ ⊢R False Example: A proof of P from {(P ∨ H) ∧ ¬H}

implies (Γ ∧ ¬ϕ) ⇒ False is valid.
Convert {(P ∨ H) ∧ ¬H} ∧ ¬P into a set of clauses
implies ¬(Γ ∧ ¬ϕ) ∨ False is valid. (CNF):
implies ¬Γ ∨ ϕ is valid.
implies Γ ⇒ ϕ is valid. {(1) (P ∨ H), (2) ¬H, (3) ¬P }

Apply the Resolution Rule:

1. (1) (P ∨ H) and (2) ¬H produces (4) P
2. (3) ¬P and (4) P produces False
Since False is produced, so ((P ∨ H) ∧ ¬H) ⇒ P is valid.

Artificial Intelligence – p.30/34 Artificial Intelligence – p.29/34

How to get CNF from (p ⇔ q) ⇔ r An Example

Theorem: If (p ⇔ q) ⇔ r and p ⇔ r, then q.
(p ⇔ q) ⇔ r ≡ ((p ⇔ q) ⇒ r) ∧ (r ⇒ (p ⇔ q))
Obtain clauses from the premises and the negation of the conclusion.
(p ⇔ q) ⇒ r ≡ ¬(p ⇔ q) ∨ r From (p ⇔ q) ⇔ r, we obtain:
(1) p ∨ q ∨ r, (2) ¬p ∨ ¬q ∨ r, (3) p ∨ ¬q ∨ ¬r, (4) ¬p ∨ q ∨ ¬r.
≡ (¬p ⇔ q) ∨ r
From p ⇔ r we obtain:
≡ ((¬p ⇒ q) ∧ (q ⇒ ¬p)) ∨ r
(5) ¬p ∨ r, (6) p ∨ ¬r
≡ ((p ∨ q) ∧ (¬q ∨ ¬p)) ∨ r
From ¬q, we obtain:
≡ (p ∨ q ∨ r) ∧ (¬q ∨ ¬p ∨ r) (7) ¬q.
Apply Resolution to the clauses:
r ⇒ (p ⇔ q) ≡ ¬r ∨ (p ⇔ q) From (7), (1): (8) p ∨ r From (10), (6): (11) p
≡ ¬r ∨ ((p ⇒ q) ∧ (q ⇒ p)) From (7), (4): (9) ¬p ∨ ¬r From (10), (9): (12) ¬p
≡ ¬r ∨ ((¬p ∨ q) ∧ (¬q ∨ p)) From (5), (8): (10) r From (11), (12): (13) 
≡ (¬p ∨ q ∨ ¬r) ∧ (¬q ∨ p¬r)
Artificial Intelligence – p.32/34 Artificial Intelligence – p.31/34
 Proof System for Satisfiability Some Resolution Strategies
The Davis-Putnam-Logemann-Loveland (DPLL) method Unit resolution: Unit resolution only.
takes a set of input clauses and converts them into an Input resolution: One of the two clauses must be an input
equivalent set of unit clauses if the set is satisfiable. clause.
Inference Rules: Set of support:
One of the two clauses must be from a
S ∪ {α ∨ β, ¬β} designed set called set of support. New resolvent are
: Unit Resolution added into the set of support.
S ∪ {α, ¬β}
Linear resolutionThe latest resolvent must be used in the
S ∪ {α ∨ β, β} current resolution.
: Unit Subsumption
S ∪ {β}
S Note: The first 3 strategies above are incomplete.
: Case Splitting The Unit resolution strategy is equivalent to the Input
S ∪ {α} or S ∪ {¬α} resolution strategy: a proof in one strategy can be
converted into a proof in the other strategy.
Artificial Intelligence – p.34/34 Artificial Intelligence – p.33/34