Knowledge Representation

and Reasoning

The AI Cycle
Almost all AI systems have the following
components in general:
Perception
Learning
Knowledge Representation and
Reasoning
Planning
Execution

The AI Cycle

Knowledge and its types

what the knowledge is?
Durkin refers to it as the Understanding of
a subject area.
A well-focused subject area is referred to
as a knowledge domain, for example,
medical domain, engineering domain,
business domain, etc..

Knowledge and its types

If we analyze the various types of
knowledge we use in every day life, we
can broadly define knowledge to be one of
the following categories:

Knowledge and its types

Procedural knowledge: Describes how to do
things, provides a set of directions of how to
perform certain tasks, e.g., how to drive a car.
Declarative knowledge: It describes objects,
rather than processes. What is known about a
situation, e.g. it is sunny today, and cherries are
red.
Meta knowledge: Knowledge about knowledge,
e.g., the knowledge that blood pressure is more
important for diagnosing a medical condition than
eye color.

Knowledge and its types

Heuristic knowledge: Rule-of-thumb, e.g. if I
start seeing shops, I am close to the market.
Heuristic knowledge is sometimes called shallow
knowledge.
Heuristic knowledge is empirical as opposed to
deterministic

Structural knowledge: Describes structures

and their relationships. e.g.
how the various parts of the car fit together to make a
car, or knowledge structures in terms of concepts, sub
concepts, and objects.

Knowledge and its types

Towards Representation
There are multiple approaches and schemes that come
to mind when we begin to think about representation.
Pictures and symbols This is how the earliest humans
represented knowledge when sophisticated linguistic
systems had not yet evolved.
Graphs and Networks
Numbers

What types of knowledge is best represented using pictures? , e.g.

can we represent the relationship between individuals in a family
using a picture?
We could use a series of pictures to store procedural knowledge,
e.g. how to boil an egg.
But we can easily see that pictures are best suited for recognition
tasks and for representing structural information.
However, pictorial representations are not very easily translated to
useful information in computers because computers cannot interpret
pictures directly with out complex reasoning.
So even though pictures are useful for human understanding,
because they provide a high level view of a concept to be obtained
readily, using them for representation in computers is not as straight
forward.

Graphs and Networks

Graphs and Networks allow relationships
between objects/entities to be incorporated, e.g.,
to show family relationships, we can use a
graph.

Graphs and Networks

We can also represent procedural
knowledge using graphs, e.g. How to start
a car?
A graph for procedural knowledge

Numbers
Numbers are an integral part of knowledge
representation used by humans.
Numbers translate easily to computer
representation.
Eventually,
as
we
know,
every
representation we use gets translated to
numbers in the computers internal
representation.

Example

Example
Using a description in words
For the family above, we could say in
words
Tariq is Monas Father
Ayesha is Monas Mother
Mona is Tariq and Ayeshas Daughter

Example
This example demonstrates the fact that
each knowledge representation scheme
has its own strengths and weaknesses.

Formal KR techniques
Facts
Facts are a basic block of knowledge (the atomic units of
knowledge).
They represent declarative knowledge (they declare knowledge
about objects).
A proposition is the statement of a fact.
Each proposition has an associated truth value. It may be either
true or false.

In AI, to represent a fact, we use a proposition and its

associated truth value, e.g.
Proposition A: It is raining
Proposition B: I have an umbrella
Proposition C: I will go to school

Formal KR techniques

Single-valued or multiple valued

Uncertain facts
Fuzzy facts
Object-Attribute-Value triplets

Single-valued or multiple valued

Facts may be single-valued or multivalued,
where each fact (attribute) can take one or more
than one values at the same time, e.g. an
individual can only have one eye color, but may
have many cars.
So the value of attribute cars may contain more
than one value.

Uncertain Facts
Sometimes we need to represent uncertain
information in facts. These facts are called
uncertain facts, e.g. it will probably be sunny
today.
We may chose to store numerical certainty
values with such facts that tell us how much
uncertainty there is in the fact.

Fuzzy Facts
Fuzzy facts are ambiguous in nature, e.g. the
book is heavy/light.
Here it is unclear what heavy means because it is a
subjective description.

Fuzzy representation is used for such facts.

While defining fuzzy facts, we use certainty
factor values to specify value of truth.
We will look at fuzzy representation in more
detail later.

Object-Attribute Value Triplets(OVT)

Object-Attribute Value Triplets(OVT) are a type of fact
composed of three parts; object, attribute and value.
Such facts are used to assert a particular property of
some object, e.g. Alis eye color is brown.
Object: Ali
Attribute: eye color
Value: brown

Ahmeds son is Ali

Object: Ahmed
Attribute: son
Value: Ali

Object-Attribute Value Triplets(OVT)

Rules
Rules are another form of knowledge
representation.
Durkin defines a rule as A knowledge
structure that relates some known
information to other information that can
be concluded or inferred to be true.

Components of a Rule
A Rule consists of two components
Antecedent or premise or the IF part
Consequent or conclusion or the THEN part

For example, we have a rule:

IF it is raining THEN I will not go to school
Premise: It is raining
Conclusion: I will not go to school.

Compound Rules
Multiple premises or antecedents may be joined
using AND (conjunctions) and OR (disjunctions),
e.g.
IF it is raining AND I have an umbrella THEN I will go
to school.
IF it is raining OR it is snowing THEN I will not go to
school.

Types of rules
Relationship
Relationship rules are used to express a direct
occurrence relationship between two events, e.g.
IF you hear a loud sound THEN the silencer is not working

Recommendation
Recommendation rules offer a recommendation on
the basis of some known information, e.g.
IF it is raining THEN bring an umbrella

Types of rules
Directive
Directive rules are like recommendations rule but they
offer a specific line of action, as opposed to the
advice of a recommendation rule, e.g.
IF it is raining AND you dont have an umbrella THEN wait for
the rain to stop

Types of rules
Variable Rule
If the same type of rule is to be applied to multiple
objects, we use variable rules, i.e. rules with
variables, e.g.
If X is a Student AND Xs GPA>3.7 THEN place X on honor
roll.

Such rules are called pattern-matching rules. The rule is

matched with known facts and different possibilities for
the variables are tested, to determine the truth of the
fact.

Types of rules
Uncertain Rules
Uncertain rules introduce uncertain facts into the
system, e.g.
IF you have never won a match THEN you will
most probably not win this time.

Meta Rules
Meta rules describe how to use other rules, e.g.
IF you are coughing AND you have chest
congestion THEN use the set of respiratory
disease rules.

Semantic networks
Semantic networks are graphs, with nodes
representing objects and arcs representing
relationships between objects.
Various types of relationships may be defined
using semantic networks.
The two most common types of relationships are
IS-A (Inheritance relation)
HAS (Ownership relation)

Semantic networks
Lets consider an example semantic
network to demonstrate how knowledge in
a semantic network can be used

Semantic networks

Semantic networks
Network Operation
To infer new information from semantic networks, we can
ask questions from nodes
Ask node vehicle: How do you travel?
This node looks at arc and replies: road
Ask node Suzuki: How do you travel?
This node does not have a link to travel therefore it
asks other nodes linked by the IS-A link
Asks node Car (because of IS-A relationship)
Asks node Vehicle (IS-A relationship)
Node Vehicle Replies: road

Problems with Semantic Networks

Semantic networks are computationally expensive at runtime.
In the worst case, we may need to traverse the entire
network and then discover that the requested info does
not exist.
They try to model human associative memory, but in the
human brain the number of neurons and links are in the
order of 1015.
It is not practical to build such a large semantic network,
hence this scheme is not feasible for this type of
problems.
Semantic networks are logically inadequate as they do
not have any equivalent quantifiers, e.g., for all, for some,
none.

Frames
Frames are data structures for representing
stereotypical knowledge of some concept or
object according to Durkin, a frame is like a
schema, as we would call it in a database
design.
They were developed from semantic networks
and later evolved into our modern-day Classes
and Objects.

Frames
For example, to represent a student, we make
use of the following frame:

The various components within the frame are

called slots, e.g. Frame Name slot.

Frames

Frames
Facets
A slot in a frame can hold more that just a value, it
consists of metadata and procedures also.

The various aspects of a slot are called facets.

They are a feature of frames that allows us to
put constraints on frames.
e.g. IF-NEEDED Facets are called when the data of a
particular slot is needed.

Similarly, IFCHANGED Facets are when the

value of a slot changes.

Logic
Just like algebra is a type of formal logic
that deals with numbers, e.g. 2+4 = 6,
propositional logic and predicate calculus
are forms of formal logic for dealing with
propositions. We will consider two basic
logic representation techniques:
Propositional Logic
Predicate Calculus

Logic
Propositional logic
A proposition is the statement of a fact. We usually
assign a symbolic variable to represent a proposition,
e.g.
p = It is raining
q = I carry an umbrella
A proposition is a sentence whose truth values may be
determined. So, each proposition has a truth value, e.g.
The proposition A rectangle has four sides is true
The proposition The world is a cube is false.

Logic
Compound statements
Different propositions may be logically related
and we can form compound statements of
propositions using logical connectives. Common
logical connectives are:

Logic
Truth Table of Binary Logical Connectives

Logic
Limitations of propositional logic
Propositions can only represent knowledge as
complete sentences, e.g. a = the balls color is blue.
Cannot analyze the internal structure of the sentence.
No quantifiers are available, e.g. for-all, there-exists
Propositional logic provides no framework for proving
statements such as:
All humans are mortal
All women are humans
Therefore, all women are mortals

This is a limitation in its representational power.

Predicate calculus
Predicate Calculus is an extension of propositional logic
that allows the structure of facts and sentences to be
defined. With predicate logic, we can use expressions like
Color( ball, blue)
This allows the relationship of sub-sentence units to be
expressed, e.g. the relationship between color, ball and
blue in the above example.
Due to its greater representational power, predicate
calculus provides a mechanism for proving statements and
can be used as a logic system for proving logical theorems.

Logic
Quantifiers
Predicate Logic
You already have done sufficient work.
Lets just have a look at an example.

PROPOSITIONAL

CALCULUS

48

PROPOSITIONAL

CALCULUS

49

PROPOSITIONAL

CALCULUS

50

PROPOSITIONAL

CALCULUS

In Propositional Calculus, each atomic

symbol denotes a proposition.
For example:it is raining

it is sunny

A proposition:
If it is raining, then it is not sunny
P Q
51

PREDICATE
CALCULUS
52

PREDICATE CALCULS
In Propositional Calculus, a single
propsitional symbol P may denote the entire
sentence it rained on Tuesday.
it rained on Tuesday =P
In Predicate Calculus, a predicate weather
may be created, that describes a relationship
between a day and the weather:
weather(tuesday,rain)
53

PREDICATE CALCULS
Predicate Calculus also allows
expressions to contain variables.
For example:
For all values of X, where X is a day of the
week, the statement weather(X,rain) is
true:
i.e. it rains every day.
54

PREDICATE CALCULS

55

PREDICATE CALCULS
Symbols are used to denote objects,
properties or relations in the world of discourse.
The use of words that suggest the symbols
intended meaning assists programmers in
understanding program code.
Thus, l ( g , k ) = like (george , kate )
Both are formally equal in predicate calculus.
But later can be a great help, for humans, in
relationship the expression represents.

indicating the

Predicate Calculus, however, concerns only with the

semantics.

formal

56

PREDICATE CALCULS
Improper Symbols. Followings are
used to form well formed expressions.
These are called Improper Symbols.
Parentheses ( )
Commas ,
Periods .

57

PREDICATE CALCULS
Predicate Calculus Term
A predicate calculus term is either

constant, variable or a function

expression.

Predicate Symbols Types

Constants
Variables
Functions
Predicates
58

PREDICATE CALCULS
Constants
Used to name specific objects or properties in
the world.
Constant symbols must begin with a
lowercase
letter.

Examples
george
tree
tall
bLUe
aRT
true and false are constants, they are
reserved as truth symbols.

59

PREDICATE CALCULS
Variables
Variable symbols represent general
classes of objects or properties in the world.
Variable symbols must begin with an
uppercase letter.

Examples
George
Tree
Tall
BLUe
ART

60

PREDICATE CALCULS
Functions
A function expression is a function symbol
followed by its argument.

functionsymbol ( argument )
Function symbol begins with a lowercase
letter.
A function expression consists of a function
constant of arity n, followed by n terms,
t1,t2,,tn, enclosed in parentheses and
separated by commas.

func ( t1,t2,t3,,tn )

61

PREDICATE CALCULS
Functions
goerge ( likes,kate ) george is a
function constant of arity 2, and there
are two terms of argument.

Examples of Functions
f(X,Y)
father(david)
price(banana)
plus(2,3)
62

PREDICATE CALCULS

63

PREDICATE CALCULS
Predicate Calculus Symbols
A predicate name a relationship between
none (zero) or more objects in the world.
Number of objects so related is the arity
(argument number) of the predicate.
Predicate symbols begin with lowercase
letters.

64

PREDICATE CALCULS
Predicate Calculus Symbols
An atomic sentence in Predicate Calculus
is a predicate of arity n followed by n terms
enclosed in parentheses and separated by
commas.

65

PREDICATE CALCULS
Structure of Atomic Sentence
likes ( goerge, kate )
Predicate of arity 2
two terms enclosed in parentheses

likes(goerge, kate, jim,tuesday )

Predicate of arity 4
4 terms enclosed in parentheses
66

PREDICATE CALCULS
Examples of Atomic Sentences
likes(george,kate)
likes(george ,susie)
friends(bill,goerge)
helps(bill,george)
likes(X,george)
likes(X,X)
friends(father(david),father(andrew))
67

PREDICATE CALCULS
Likes, friends, help are predicate symbols.
Predicates with the same names but
different
arities are considered different.
Bill, george, kate etc are constant symbols
and represent objects in the problem domain.
The arguments to a predicate are terms and
may also include variables or function
expressions. For` example:friends(father(david),father(andrew))
68

PREDICATE CALCULS
Atomic sentences are also called
atomic
expressions, atoms or
propositions.
Atomic sentences may be combined
with connectives:

, , , ,
A variable in the sentence refers to
unspecified objects in the domain.

69

PREDICATE CALCULS
Predicate calculus includes two symbols that
defines the meaning of a sentence containing a
variable. These are variable quantifiers:

Universal quantifier : Indicate that

sentence is true for all values of variable in the
domain.

Existential quantifier -- : Indicate that

sentence is true for some values of variable in
the domain.
70

PREDICATE CALCULS
A quantifier is followed by a variable and a
sentence, such that:
Y friends(Y,peter): A few elements of
the
domain are friends of
Peter.
X likes(X,ice_cream): All elements of
the
domain like ice cream.
71

PREDICATE CALCULS
A few examples of sentences represented in
predicate calculus:
If it doesnt rain tomorrow, Aslam will go to the
mountains.
weather (rain , tomorrow) go (aslam , mountains)
All basketball players are tall.
X (basketball_player (X) tall(X))
72

PREDICATE CALCULS
Some people like bananas.
X (person (X) likes ( X, bananas)).
Nobody likes taxes.
X likes (X , taxes)

73

PREDICATE CALCULUS SENTENCES

Every atomic sentence is a sentence:
If s is a sentence,

so is its negation s.

If s1 and s2 are sentences, then so is their

conjunctions1 s2.
If s1 and s2 are sentences, then so is their
disjunction s1 s2.
If s1 and s2 are sentences, then so is their
implication s1 s2.
74

PREDICATE CALCULUS SENTENCES

Every atomic sentence is a sentence:
If s1 and s2 are sentences, then so is their
equivalence s1 s2.
If X is a variable and s is a sentence,
then X s is a
sentence.
If X is a variable and s is a sentence,
then X s is a
sentence.
75

EXAMPLES OF WELL FORMED

PREDICATE SENTENCES
Let times and plus be function symbols of arity
2.
Let equal and foo be predicate symbols of arity
2 and 3 respectively.
plus (2 , 3) is a function. Not an atomic
sentence.
76
equal (plus (2 , 3) , five) Is an atomic sentence.

EXAMPLES OF WELL FORMED

PREDICATE SENTENCES
equal (plus (2 , 3) , seven) Is an atomic
sentence.
Though the interpretation value of this
sentence is false. But well formedness and
truth values are independent issues.

77

EXAMPLES OF WELL FORMED

PREDICATE SENTENCES
X foo (X, two, plus (2, 3)) equal (plus (2, 3), five)
is a sentence because both the conjuncts are
sentences.
(foo (two, two, plus (2, 3))) (equal (plus (2, 3), five) =
true)
Is a sentence because all its components are
sentences connected by logical operators.
78

SEMANTICS FOR THE PREDICATE

CALCULUS
Predicate calculus semantics provide a formal
basis for determining the truth values of
expressions.
The truth of expressions depends on the
mapping of constants, variables, predicates,
and functions on to the objects and relations
in the domain of discourse.
79

SEMANTICS FOR THE PREDICATE

CALCULUS
For example:
Information about a person George and his
friends Kate and Susie may be expressed
by:

friends(george,susie)
friends(george,kate)

80

SEMANTICS FOR THE PREDICATE

CALCULUS
If it is true that George is a friend of Kate and
Susie, then each expression would have truth
value (assignment) T.
If George is friend of Kate but not of Susie,
then, friends (george, kate) has the truth value
T and, friends (george, susie) has the truth
value F.

81

INTERPRETATION
Let the domain D be a nonempty set. An
interpretation over D is an assignment of the
entities of D to each of the constant, variable,
predicate and function symbols of a predicate
calculus expression, such that:
Each constant is assigned an element of D.
Each variable is assigned to a nonempty subset of
D. These are the allowable substitutions for that
variable.
82

INTERPRETATION
Each function f of arity m is defined on m
arguments of D. defines a mapping from Dm
into D.
Each predicate p of arity n is defined on n
arguments from D defines a mapping from Dn
into { T, F}.

83

INTERPRETATION
Sentence:
friends (father_of (david), father_of (andrew) )
Functions:
father_of(david)

function evaluation : george

father_of(andrew)

function evaluation : allen

Semantics of expression:
friends(george, allen)
84

INFERENCE RULES
The ability to infer new correct expressions
from a set of true assertions is an
important feature of predicate calculus.
These new expressions are correct in that
they are consistent with all previous
interpretations of the original set of
expressions.
85

INFERENCE RULES
An interpretation that makes a sentence
true is said to satisfy that sentence.
An interpretation that satisfy every
member

of a set of expressions is said to

satisfy

the set.

86

INFERENCE RULES
An expression X logically follows from a set
of predicate calculus expressions S if every
interpretation that satisfy S also satisfy X.
The function of logical inference is to produce
new sentences that logically follow a set of
predicate calculus sentences.

87

INFERENCE RULES
An inference rule is essentially a mechanical
means of producing new predicate calculus
sentences from other sentences.
S 1: X human (X) mortal (X)
I: all human die.

If Socrates is substituted for X

S 2: human(socrates) mortal (socrates).
S2, an expression logically following from S1,
also satisfied by I.
88

DEFINITIONS
SATISFY, MODEL, VALID, INCONSISTENT
For a predicate calculus expression X and an
interpretation I:
If X has a value of T under I and a
particular variable assignment, then I is
said to satisfy X.
If I satisfies X for all variable assignments,
then I is a model of X.
89

DEFINITIONS
SATISFY, MODEL, VALID, INCONSISTENT
For a predicate calculus expression X and an
interpretation I:
X is satisfiable if and only if there exist an
interpretation and variable assignment that
satisfy it. Otherwise, it is unsatisfiable.

90

DEFINITIONS
SATISFY, MODEL, VALID, INCONSISTENT
A set of expressions is satisfiable if and only if
there exist an interpretation and variable
assignment that satisfy every element.
If a set of expression is not satisfiable, it is said
to be inconsistent.
If X has a value T for all possible interpretations,
X is said to be valid.
91

DEFINITIONS
LOGICALLY FOLLOWS, SOUND AND
COMPLETE

A predicate calculus expression X logically

follows from a set S of expressions, if every

interpretation and variable assignment that
satisfies S also satisfies X.
An inference rule is sound if every predicate
calculus expression produced by the rule from a
set S of expressions also logically follows S.
92

DEFINITIONS
LOGICALLY FOLLOWS, SOUND AND
COMPLETE
An inference rule is complete if, given a
set S of predicate calculus expressions, the
rule can infer every expression that
logically follows from S.

93

INFERENCE RULES
If the sentences P and P Q are known to be
true, then inference rule Modus Ponens lets us
infer Q.
Under the inference rule Modus Tollen,
if P Q is known to be false, we can infer P.

94

INFERENCE RULES
AND Elimination allows us to infer the truth of
either of the conjuncts from the truth of a
conjunctive sentence. For instance, P Q lets us
conclude P and Q are true.

95

INFERENCE RULES
AND Introduction lets us infer the truth of a
conjunction from the truth of its conjuncts, for
instance, if P and Q are true, then P Q is true.

96

UNIFICATION
To apply inference rules such as modus
ponens, an inference system must be able to
determine when two expressions are the
SAME or MATCHING.
Unification is an algorithm for determining
the substitutions needed to replace the
variables, in order to make two predicate
calculus expressions MATCH.
97

UNIFICATION
Unification and inference rules such as modus
ponens allow us to make inferences on a set of
logical assertions. To do this the logical database
must be expressed in an appropriate form: All variables be universally quantified.
Existentially quantified variables may be eliminated from
sentences in the database by replacing them with the
constants that make the sentence true.

98

UNIFICATION
A variable may be replaced by any term: Other variables.
Function expressions.
For example a few instances of the expression
foo (X, a, goo (Y)):Instances
Substitutions or Unifications
foo (fred, a, goo (Z))
{fred / X, Z / Y}
foo(W, a, goo(jack))
{W / X, jack / Y}
foo (Z, a, goo(moo(Z)))
{Z / X, moo (Z) / Y}
99

UNIFICATION
A variable is said to be BOUND to the value
substituted for it. Therefore substitutions are also
referred to be BINDINGS.

100

Reasoning
Now that we have looked at knowledge representation,
we will look at mechanisms to reason on the knowledge
once we have represented it using some logical scheme.
Reasoning is the process of deriving logical conclusions
from given facts.
Durkin defines reasoning as the process of working with
knowledge, facts and problem solving strategies to draw
conclusions.

Types of reasoning

Deductive reasoning
Inductive reasoning
Abductive reasoning
Analogical reasoning
Common-sense reasoning
Non-Monotonic reasoning

Deductive reasoning
Deductive reasoning, as the name implies, is based on
deducing new information from logically related known
information.
A deductive argument offers assertions that lead
automatically to a conclusion, e.g.
If there is dry wood, oxygen and a spark, there will be
a fire
Given: There is dry wood, oxygen and a spark
We can deduce: There will be a fire.

All men are mortal. Socrates is a man.

We can deduce: Socrates is mortal

Inductive Reasoning

Inductive reasoning is based on forming, or

generalization from a limited set of observations, e.g.

inducing

Observation: All the crows that I have seen in my life are black.
Conclusion: All crows are black

Comparison of deductive and inductive reasoning

We can compare deductive and inductive reasoning using an example.
We conclude what will happen when we let a ball go using both each
type of reasoning in turn
The inductive reasoning is as follows: By experience, every time I have
let a ball go, it falls downwards. Therefore, I conclude that the next time
I let a ball go, it will also come down.
The deductive reasoning is as follows: I know Newton's Laws. So I
conclude that if I let a ball go, it will certainly fall downwards.

Thus the essential difference is that inductive reasoning is based on

experience while deductive reasoning is based on rules, hence the
latter will always be correct.

Abductive reasoning
Deduction is exact in the sense that deductions follow in
a logically provable way from the axioms. Abduction is a
form of deduction that allows for plausible inference, i.e.
the conclusion might be wrong, e.g.
Implication: She carries an umbrella if it is raining
Axiom: she is carrying an umbrella
Conclusion: It is raining

This conclusion might be false, because there could be

other reasons that she is carrying an umbrella, e.g. she
might be carrying it to protect herself from the sun.

Analogical reasoning
Analogical reasoning works by drawing
analogies between two situations, looking
for similarities and differences, e.g.
when you say driving a truck is just like driving
a car, by analogy you know that there are
some similarities in the driving mechanism,
but you also know that there are certain other
distinct characteristics of each.

Common-sense reasoning
Common-sense reasoning is an informal form of
reasoning that uses rules gained through experience or
what we call rules-of-thumb.
It operates on heuristic knowledge and heuristic rules.

Non-Monotonic reasoning
Non-Monotonic reasoning is used when the facts of the
case are likely to change after some time, e.g.
Rule: IF the wind blows
THEN the curtains sway

When the wind stops blowing, the curtains should sway

no longer. However, if we use monotonic reasoning, this
would not happen.
The fact that the curtains are swaying would be retained
even after the wind stopped blowing. In non monotonic
reasoning, we have a truth maintenance system. It
keeps track of what caused a fact to become true. If the
cause is removed, that fact is removed (retracted) also.