2/18/2024

KNOWLEDGE
REPRESENTATION Introduction

Dr. Sumit Srivastava

Dept. of CSE, BIT Mesra Ranchi
Email:- [email protected]

Sumit Srivastav @ BIT Mesra

1 2

Introduction Introduction

There are three factors which are put into the machine, which
 Human beings are good at understanding, reasoning and
makes it valuable:
interpreting knowledge.
 Knowledge: The information related to the environment is
 And using this knowledge, they are able to perform various
stored in the machine.
actions in the real world. But how do machines perform
the same?  Reasoning: The ability of the machine to understand the stored
knowledge.

 Intelligence: The ability of the machine to make decisions on

the basis of the stored information.

3 4

Relation between Knowledge & Intelligence Introduction

In this example, there is one decision-maker whose actions are justified by sensing the
environment and using knowledge.
But, if we remove the knowledge part here, it will not be able to display any intelligent
behavior

5 6

1
 2/18/2024

Knowledge Representation What is Knowledge Representation?

 Knowledge is cognizance or understanding expanded by  Knowledge Representation in AI describes the

experiences of facts, data, and circumstances. representation of knowledge.

 Humans are best at understanding, reasoning, and  Basically, it is a study of how the beliefs, intentions, and
interpreting knowledge. But how machines will be able to judgments of an intelligent agent can be expressed suitably
do all these comes under knowledge representation and for automated reasoning.
reasoning.
 One of the primary purposes of Knowledge Representation
includes modeling intelligent behavior for an agent.

7 8

What is Knowledge Representation? Knowledge Representation

 Knowledge Representation and Reasoning (KR, KRR)

represents information from the real world for a computer to
 KNOWLEDGE:- It refer to the fact of knowing

understand and then utilize this knowledge to solve complex something that has been gained through an experience and

real-life problems like communicating with human beings in learning.

natural language.  REPRESENTATION:- It refer to the process of

 Knowledge representation in AI is not just about storing data representing what has been gained from the knowledge.

in a database, it allows a machine to learn from that

knowledge and behave intelligently like a human being.

9 10

What Knowledge Needs to be Represented? What to Represent

 Object: Info about the objects in our world, like knowing
 The different kinds of knowledge that need to be represented is
buses need drivers or that guitars have strings.
 Object
 Events: Events are the actions which occur in our world.
 Events
 Performance  Performance: It describe behavior which involves knowledge
 Meta-knowledge about how to do things. (Understanding how people behave in
 Facts different situations.)

 Knowledge base  Meta-knowledge: It is knowledge about what we know.

11 12

2
 2/18/2024

What to Represent Types of Knowledge

 Facts: Facts are the truths about the real world and what we
represent.

 Knowledge-Base: The central component of the knowledge-

based agents is the knowledge base. It is represented as KB. It
refers to a set of information about any discipline, field, etc.
For example, a knowledge base on road construction.

13 14

Types of Knowledge Properties of Knowledge Representation

1. Declarative Knowledge (The “What” Knowledge) • Representational Accuracy: It should represent all kinds of
• It’s all about facts and concepts, helping describe things in required knowledge.
simple terms.
2. Structural Knowledge (The “How Things Relate” Knowledge) • Inferential Adequacy: The system should be flexible, allowing
• This knowledge helps AI understand relationships between it to adjust old knowledge to fit new information.
concepts and objects, aiding problem-solving.
3. Procedural Knowledge (The “How-To” Knowledge) • Inferential Efficiency: The representation system cannot
• This is like a manual for tasks, with specific rules and strategies accommodate new knowledge in the presence of the old
to follow. knowledge, but it can add this knowledge efficiently and in a
4. Meta Knowledge (What We Already Know) seamless manner.
• It’s knowledge about knowledge, including categories, plans, and
past learning. • Acquisitional Efficiency: The ability to acquire new knowledge
5. Heuristic Knowledge (Learning from Experience) easily using automatic methods.
• This type helps AI make decisions based on past experiences,
like using old techniques to solve new problems.

15 16

AI Knowledge Cycle AI Knowledge Cycle

 An AI system has the following components for displaying
intelligent behavior:

 Perception

 Learning

 Knowledge Representation and Reasoning

 Planning

 Execution

17 18

3
 2/18/2024

Knowledge Representation Techniques

Knowledge Representation
Techniques

19 20

Logical Representation Logical Representation

 Logical representation is a language with some definite rules
which deal with propositions and has no ambiguity in
representation.
 It represents a conclusion based on various conditions and
lays down some important communication rules.
 Also, it consists of precisely defined syntax and semantics
which supports the sound inference.
 There are two type of logical representation:
 Each sentence can be translated into logics using syntax and 1. Propositional Logic
semantics. 2. Predict Logic

21 22

Logical Representation Semantic Network Representation

Advantages of logical representation:  Semantic networks work as an alternative of predicate logic for
 Logical representation enables us to do logical reasoning. knowledge representation. In Semantic networks, you can
 Logical representation is the basis for the programming represent your knowledge in the form of graphical networks.
languages.  This network consists of nodes representing objects and arcs
Disadvantages of logical Representation: which describe the relationship between those objects. Also, it
 Logical representations have some restrictions and are categorizes the object in different forms and links those objects.
challenging to work with.  This representation consist of two types of relations:
 Logical representation technique may not be very natural, IS-A relation (Inheritance)
Kind-of-relation
and inference may not be so efficient.

23 24

4
 2/18/2024

Semantic Network Representation Semantic Network Representation

Advantages:
Example:
 Semantic networks are a natural representation of knowledge.
Statements:
 Also, it conveys meaning in a transparent manner.
Jerry is a cat.  These networks are simple and easy to understand.
Jerry is a mammal Disadvantages:
Jerry is owned by Priya.  Semantic networks take more computational time at runtime.
Jerry is brown colored.  Also, these are inadequate as they do not have any equivalent
All Mammals are animal. quantifiers.
 These networks are not intelligent and depend on the creator of the
system.

25 26

Frame Representation Frame Representation

Advantages:
 A frame is a record like structure that consists of a collection of
 It makes the programming easier by grouping the related data.
attributes and values to describe an entity in the world.
 Frame representation is easy to understand and visualize.
 These are the AI data structure that divides knowledge into
 It is very easy to add slots for new attributes and relations.
substructures by representing stereotypes situations.
 Also, it is easy to include default data and search for missing values.
 Basically, it consists of a collection of slots and slot values of any
Disadvantages:
type and size.
 In frame system inference, the mechanism cannot be easily processed.
 Slots have names and values which are called facets.
 The inference mechanism cannot be smoothly proceeded by frame
representation.
 It has a very generalized approach.

27 28

Production Rules Production Rules

 A frame is a record like structure that consists of a collection  It is based on pair of (condition, action) that is “If condition
then action”.
of attributes and values to describe an entity in the world.
 It has mainly three parts:
 These are the AI data structure that divides knowledge into  Set of production rules
substructures by representing stereotypes situations.  Working Memory
 The recognize-act-cyle
 Basically, it consists of a collection of slots and slot values  Condition part- Determines which rule may be applied to a
of any type and size. problem.
 Action part- Carries out the associated problem-solving
 Slots have names and values which are called facets. steps.

This complete process is known as recognize-act cycle

29 30

5
 2/18/2024

Production Rules Production Rules

 Example: Advantages:
 The production rules are expressed in natural language.
IF (at bus stop AND bus arrives)
THEN action (get into the bus)  The production rules are highly modular and can be easily
IF (on the bus AND paid AND empty seat) removed or modified.
THEN action (sit down).
IF (on bus AND unpaid) Disadvantages:
THEN action (pay charges).  It does not exhibit any learning capabilities and does not
IF (bus arrives at destination)
THEN action (get down from the bus). store the result of the problem for future uses.
 During the execution of the program, many rules may be
active. Thus, rule-based production systems are inefficient.

31 32