Dr.

Shahnawaz Qureshi
 Humans know things …
 the knowledge helps them to do various tasks.

 The knowledge has been achieved

◦ not by purely reflex mechanisms
◦ but by the processes of reasoning
 In AI, the example is knowladge-based agent which
contains set of sentences referred as knowledge-base.

3
For a generic knowledge-based agent:

 A percept is given to the agent.

 The agent adds the percept to its knowledge base.

 Perform best action according to the knowledge base.

 Tells the knowledge base that it has in fact taken that

action.

4
constructs a sentence asserting that the agent
perceived the given percept at time t
constructs a sentence that asks what action
should be done at time t
constructs a sentence that the chosen action
was executed at time t
5
6
7
The PEAS description for Wumpus World:
Performance measure:
 +1000 for climbing out of the cave with the gold,
 –1000 for falling into a pit or being eaten by the Wumpus,
 –1 for each action taken
 –10 for using up the arrow
Environment:
 A 4×4 grid of rooms. The agent starts in the square labelled
[1,1], facing to the right.
The game ends either when the agent dies or when the
agent climbs out of the cave.

8
The PEAS description for Wumpus World:
Actuators:
 The agent can move Forward, TurnLeft by 90◦, TurnRight by
90◦, grab, shoot
Sensors:
 The square adjacent directly (not diagonally) to the square
containing Wumpus, the agent will perceive a Stench.
 The squares adjacent to a pit, the agent will perceive a
Breeze.
 The square with gold, the agent will perceive a Glitter.
 An agent walks into a wall, it will perceive a Bump.
 When the Wumpus is killed, it emits a woeful Scream.

9
10
B

11
P

12
S
P

13
W
S OK
P

14
W
S OK
P

15
 BGS OK
W
S OK OK
P

16
17
 The knowledge bases consist of sentences.

How to represent these sentances?

 Logic, a formal language, is the solution --- a way of

manipulating expressions in the language.

 Logic has
◦ Syntax
◦ Semantics

19
Syntax:
What expressions are legal --- what are allowed to write
down.
 The notion of syntax is clear enough with the example:
“𝑥 + 𝑦 = 4” is a well-formed sentence, whereas
“𝑥4𝑦+ =” is not.
Semantics:
What legal expression means --- meaning of sentences
 the sentence “𝑥 + 𝑦 = 4” is true in a world where 𝑥 is 2
and 𝑦 is 2, but false in a world where 𝑥 is 1 and 𝑦 is 1.
 Syntax is a form and semantics is the content.

20
Semantics:
 The semantics defines the truth of each sentence with
respect to each possible world.

 The term model can be used in place of “possible

world.”

 If a sentence 𝜶 is true in model 𝒎, we say that 𝒎

satisfies 𝜶 or sometimes 𝒎 is a model of 𝜶.

 The notation 𝑴(𝜶) --- the set of all models of 𝜶.

21
Entailment:
 means that one thing follows from another:

𝜶╞𝜷

 if and only if, in every model in which 𝜶 is true, 𝜷 is

also true. We can write

 The notation ⊆ means that: if 𝜶 ╞ 𝜷, then 𝜶 is a

stronger assertion than 𝜷

22
 We say 𝒎 is a model of sentence 𝜶 if 𝜶 is true in 𝒎

 𝑴(𝜶) is the set of all models of 𝜶

Then 𝑲𝑩 ╞ 𝜶 iff 𝑴(𝑲𝑩) ⊆ 𝑴(𝜶)

Example:
 The sentence 𝒙 = 𝟎 entails the sentence 𝒙𝒚 = 𝟎
◦ In any model where 𝒙 is zero, it is the case that 𝒙𝒚 is zero
(regardless of the value of 𝒚)

23
 Situation after detecting
nothing in [1,1], moving
right, breeze in [1,2]

 Consider possible models for

KB assuming only pits

 3 Boolean choices  8
possible models

24
3 Boolean choices  8 possible models
regardless of wumpus-world rules
25
KB = wumpus-world rules + observations

26
𝑲𝑩 = wumpus-world rules + observations
𝜶𝟏 = "[1,2] is safe", 𝑲𝑩 ╞ 𝜶𝟏 , proved by model checking
27
𝑲𝑩 = wumpus-world rules + observations
𝜶𝟐 = "[2,2] is safe", 𝑲𝑩 ╞ 𝜶𝟐
28
 If an inference algorithm 𝒊 can derive 𝜶 from 𝑲𝑩, we write

 which is pronounced “𝜶 is derived from 𝑲𝑩 by 𝒊” or

“𝒊 derives 𝜶 from 𝑲𝑩.”
Soundness:
 An inference algorithm that derives only entailed
sentences is called sound or truth preserving.
 Soundness is a highly desirable property.
Completeness:
 An inference algorithm is complete if it can derive any
sentence that is entailed.

29
 We'll look at two kinds of logic:

Propositional Logic
which is relatively simple.

First-order Logic
which is more complicated.

30
 The syntax of propositional logic defines the allowable
sentences.

What are the sentances?

 Sentance are well formed formulas

 True and False are sentances
 Propositional variables are sentences. P, Q, R, S etc.

32
 The atomic sentences consist of a single proposition
symbol.
 Each such symbol stands for a proposition that can be
True or False.
 The complex sentences are constructed from simpler
sentences, using parentheses and logical connectives.
 There are five connectives in common use:
￢ (not), ∧ (and), ∨ (or),
⇒ (implies), ⇔ (if and only if)

33
 ￢ (not) A sentence such as ￢W1,3 is called the
negation of W1,3.
◦ A literal is either an atomic sentence (a positive literal) or a
negated atomic sentence (a negative literal).
 ∧ (and) A sentence whose main connective is ∧, such
as W1,3 ∧ P3,1, is called a conjunction; its parts are the
conjuncts.
 ∨ (or) A sentence using ∨, such as (W1,3 ∧ P3,1) ∨ W2,2,
is a disjunction of the disjuncts (W1,3 ∧ P3,1) and W2,2.

34
 ⇒ (implies) A sentence such as (W1,3∧P3,1) ⇒ ￢W2,2 is
called an implication (or conditional). The premise or
antecedent is (W1,3 ∧P3,1).
 Implications are also known as rules or if–then
statements.
 The implication symbol is sometimes written as ⊃ or →
or ⇒.
 ⇔ (if and only if) The sentence W1,3 ⇔ ￢W2,2 is a bi-
conditional. Sometime it is written as ≡.

35
36
37
Precedence Example:

38
 The semantics defines the rules for determining the
truth value of a sentence with respect to a particular
model.
 In propositional logic, a model simply fixes the truth
value—true or false—for every proposition symbol
For example:
 If the sentences in the knowledge base make use of
the proposition symbols P1,2, P2,2, and P3,1, then one
possible model is:

39
 The semantics for propositional logic must specify how
to compute the truth value of any sentence, given a
model.
For Atomic sentences:
 True is true in every model and False is false in every
model.
 The truth value of every other proposition symbol must
be specified directly in the model.
◦ For example, in the model 𝒎𝟏 given earlier, P1,2 is false.

40
For complex sentences
 We have five rules, which hold for any sub-sentences 𝑷
and 𝑸 in any model 𝒎

41
 The propositional logic does not require any relation
of causation or relevance between P and Q.
◦ For example, the sentence “5 is odd implies Tokyo is the
capital of Japan” is a true sentence of propositional logic,
even though it is not a well-formed English sentence.

 In case of implication, any implication is true

whenever its antecedent is false.
◦ For example, “5 is even implies Sam is smart” is true,
regardless of whether Sam is smart or not.

42
43
 With propositional logic, we can construct a knowledge
base for the Wumpus world.

For Example:

44
 There is no pit in [1,1]:

 A square is breezy if and only if there is a pit in a

neighbouring square. This has to be stated for each
square; for now, we include just the relevant squares:

45
46
 Artificial Intelligence, A Modern Approach
Stuart J. Russell and Peter Norvig
◦ Chapter 7.

47
48