Humans excel at understanding, reasoning, and interpreting knowledge.

They use what they

know to make decisions and take actions in the real world. For machines to do something
similar, they rely on knowledge representation and reasoning. Here's what that means:

What is Knowledge Representation and Reasoning (KRR)?

 Knowledge Representation and Reasoning (KRR) is a key part of Artificial

Intelligence (AI) that deals with how AI systems think and how that thinking leads to
intelligent actions.
 It allows computers to represent information about the world and use this knowledge to
solve problems like diagnosing medical conditions or having conversations in natural
language.
 Unlike just storing data, KRR helps machines learn from knowledge and experiences,
enabling them to act intelligently, similar to how humans behave.

Types of Knowledge to Represent in AI:

1. Objects: Information about objects in the world (e.g., "guitars have strings" or "trumpets
are brass instruments").
2. Events: Actions or occurrences that happen in the world.
3. Performance: Knowledge about how to perform tasks or actions.
4. Meta-Knowledge: Knowledge about what we already know.
5. Facts: Truths about the world that need to be represented.
6. Knowledge Base (KB): The core of AI knowledge systems, a Knowledge Base is a
collection of information, often stored as sentences (not necessarily in natural language).
This helps AI systems make sense of and use the knowledge to make decisions.

In AI, KRR is essential for machines to act intelligently and interact with the world in a
meaningful way.

Types of Knowledge:

1. Declarative Knowledge:
o Knowledge about facts and objects.
o Expressed in declarative sentences (e.g., "Cats are animals").
o Simpler than procedural knowledge.
2. Structural Knowledge:
o Basic knowledge for problem-solving.
o Describes relationships between concepts (e.g., "a dog is a type of mammal").
3. Procedural Knowledge:
o Known as imperative knowledge, it defines how to perform tasks.
o Includes rules, strategies, and procedures.
o Task-specific and directly applicable.
4. Meta-Knowledge:
 o Knowledge about knowledge itself (e.g., knowing what you know and don’t
know).
5. Heuristic Knowledge:
o Expertise-based rules of thumb derived from experience.
o Helpful for problem-solving but not guaranteed to be perfect.

Ways of Knowledge Representation:

1. Logical Representation:
o Uses logic (syntax and semantics) to represent knowledge.
o Propositional and Predicate logic allow us to make conclusions based on given
conditions.
o Advantages: It enables logical reasoning and can be used in programming for
precise outcomes.
o Disadvantages: It can be restrictive, meaning it doesn't always work in every
situation, and may be inefficient at times.
2. Semantic Network Representation:
o Represents knowledge by using nodes (objects) and edges (relationships), which
show how things are connected.
o Advantages: It's easy to understand and visualize, making it helpful for
representing complex relationships.
o Disadvantages: It can be time-consuming to build, and it lacks standardized ways
to label the connections between nodes.
3. Frame Representation:
o Organizes related information using "slots" (categories) and "values" (details),
much like how modern programming languages use classes and objects.
o Advantages: It's easy to organize and modify information when necessary.
o Disadvantages: The methods for drawing conclusions (inference) from this
structure are not always efficient.
4. Production Rules:
o Uses "if-then" rules (condition, action) for decision-making, much like following
a set of instructions.
o Advantages: These rules are easy to understand since they are expressed in
natural language and are modular, meaning they can be used in different
situations.
o Disadvantages: They don’t have the ability to learn from new information and
can sometimes be inefficient during execution.

Each of these methods helps machines to understand, reason, and act based on the knowledge
they receive, aiming to mimic how humans process information.
Script:

A script is a structured way of describing a typical sequence of events that happen in a certain
situation. Scripts are used in systems that understand natural language to organize knowledge
about common situations, such as going to the movies, eating at a restaurant, shopping, or
visiting a doctor.

In essence, a script outlines a set of expected events that usually follow one another.

Components of a Script:

 Entry condition: These are the basic conditions that must be met before the events in the script
can happen.
 Results: Conditions that will be true after the events have occurred.
 Props: The objects involved in the events, like tools or materials.
 Roles: The actions performed by the people or participants involved.
 Track: Different variations of the script, where different versions may share the same
components.
 Scenes: The sequence of events that take place.

Special Symbols Used in Scripts:

Scripts use symbols to represent actions. Some common symbols are:

 ATRANS: Transfer a relationship (e.g., give)

 PTRANS: Transfer the physical location of an object (e.g., go)
 PROPEL: Apply physical force to an object (e.g., push)
 MOVE: Move a part of the body (e.g., kick)
 GRASP: Grab an object (e.g., hold)
 INGEST: Take an object into the body (e.g., eat, drink)
 EXPEL: Expel something from the body (e.g., cry)
 MTRANS: Transfer mental information (e.g., tell)
 MBUILD: Create new information mentally (e.g., decide)
 CONC: Think about an idea (e.g., think)
 SPEAK: Produce sound (e.g., say)
 ATTEND: Focus senses (e.g., listen)

Example: Script for Withdrawing Money at a Bank

 Script: Withdraw money

 Track: Bank
 Props: Money, Counter, Form, Token
 Roles:
o P = Customer
o E = Employee
o C = Cashier
  Entry conditions:
o The customer (P) has little or no money.
o The bank is open.
 Results:
o The customer (P) has more money.

Scenes:

1. Entering the Bank:

o Customer (P) walks into the bank.
o Customer (P) looks at the employee (E).
o Customer (P) approaches the employee (E).

2. Filling Out the Form:

o Customer (P) signals to the employee (E).
o Employee (E) gives the form to the customer (P).
o Customer (P) fills out the form.
o Customer (P) gives the form back to the employee (E).

3. Withdrawing Money:
o Customer (P) looks at the counter.
o Customer (P) queues at the counter.
o Customer (P) gives the token to the cashier (C).
o Cashier (C) gives money to the customer (P).

4. Exiting the Bank:

o Customer (P) leaves the bank.

Advantages of Scripts:

 They can predict future events based on previous knowledge.

 A single, clear understanding can be formed from multiple observations.

Disadvantages of Scripts:

 Scripts are less flexible than frames.

 They may not be useful for representing all types of knowledge.