Unit 3

Que 1. Analyze knowledge representation in artificial intelligence?

Ans.

What is it?

Knowledge Representation (KR) is how we teach an AI what it needs to know. It’s not just saving
data like a hard drive; it's organizing that data so the AI can use it to "think" and make decisions,
just like a human uses their memory to solve problems.

Key things it must do:

1. Represent Objects: It needs to know what things are (e.g., a "cat") and their features
(e.g., "has whiskers").

2. Represent Events: It needs to understand actions (e.g., "The cat drank milk").

3. Reasoning: The most important part. If it knows "Cats hate water" and "Fluffy is a cat," it
should be able to figure out "Fluffy hates water" without being explicitly told.

Qualities of a good system:

• Completeness: Can it describe everything in its specific area?

• Speed: Can it find answers quickly?

• Learning: Can we easily add new facts to it?

Que 2. List and explain the major issues in knowledge representation.

Ans.

It is very hard to put real-world knowledge into a computer. Here are the main problems:

1. The Detail Problem (Granularity): How much detail is needed?

o If you just say "John ate," it might be too vague.

o If you say "John picked up the fork, opened his mouth, chewed 5 times..." it is too
much useless detail. Finding the middle ground is hard.

2. The "Maybe" Problem (Uncertainty): Computers like "Yes" or "No". The real world has
"maybe," "probably," and "sometimes." Traditional logic struggles with this.

3. The Change Problem (Frame Problem): When you do one action (like close a door),
millions of other things don't change (the wall color, the weather). A computer has to be
told that these things didn't change, which is overwhelming.

4. Choosing the Right Format: Some knowledge fits better in a list, some in a graph, and
some in strict logical rules. Choosing the wrong format makes thinking slow.
Que 3. How does uncertainty affect knowledge representation systems?

Ans.

What is Uncertainty?

It means not being 100% sure about information. It happens because data might be missing,
unreliable, or just vague (like saying someone is "tall"—how tall is exactly "tall"?).

How it affects AI:

Standard AI logic uses simple True/False rules. Uncertainty breaks these rules.

• If an AI is 80% sure a patient has the flu, it can't just say "True" (they have it) or "False"
(they don't).

• It forces us to use different systems, like Probability (math for measuring chance) or
Fuzzy Logic (which allows things to be "partly true").

• Without handling uncertainty, an AI would freeze up or make bad errors whenever it

Que 4. Enlist the components of First Order Logic.

Ans.

First-Order Logic is a formal language for writing down facts clearly. Its parts are:

1. Constants: Actual names of things (e.g., Batman, Paris).

2. Variables: Placeholders for general things (e.g., x, y - like in algebra).

3. Predicates: The facts or relationships about things (e.g., Is Rich (Batman) works like
"Batman is rich").

4. Connectives: Words that join parts of sentences:

o AND (∧)

o OR (∨)

o NOT (¬)

o IF-THEN (→)

5. Quantifiers: Words that tell you “How many":

o ∀ (means ALL).

o ∃ (means SOME or AT LEAST ONE).

6. Equality (=): Says two different names belong to the same object.
Que 5. Explain how facts and rules are represented in First-Order Logic?

Ans.

In FOL, we split knowledge into two simple buckets: concrete facts and general rules.

1. Facts (Atomic Sentences):

These are simple truths about one specific thing.

• Format: Relationship (Object)

• Example: "Tweety is a Bird" is written as Bird (Tweety).

• It is a direct assertion. It is always True in the knowledge base.

2. Rules (Complex Sentences):

These are general patterns that apply to many things at once. They usually use variables (like x)
and "If-Then" arrows.

• Format: "For all x, IF x is this, THEN x is that."

• Example: "All birds can fly" is written as ∀x (Bird(x) → CanFly(x)).

• This rule lets the AI figure out that since Tweety is a bird (Fact 1), Tweety must be able to
fly.

Que 6. What is predicate logic? How is it used in AI?

Ans.

What is it? Predicate logic is a mathematical way of writing sentences so a computer can
understand them perfectly. It is better than basic logic because it can break a sentence down
into objects (like 'John') and relationships (like 'loves Mary'), rather than just seeing the whole
sentence as one big blob.

How is it used in AI?

1. Storing Knowledge: It is the standard way to write down clear, unbreakable rules for an
AI's memory.

2. Inference (Thinking): AI uses strict mathematical rules to combine predicate logic

3. Language Processing: It helps translate messy human language into neat logical
formulas the computer can actually work with.
Que 7. How are universal and existential quantifiers used in predicate logic?

Ans.

Quantifiers are symbols that let us talk about groups of things instead of just individual items.

1. Universal Quantifier (∀ - The "ForAll" symbol)

• Used for: Rules that have NO exceptions.

• Meaning: "Every single one of these is like this."

• How to use: It almost always goes with "IF-THEN" (→).

• Example: ∀x (King(x) → Person(x)) means "For absolutely every $x$, IF it is a King, THEN it
is a Person."

2. Existential Quantifier (∃ - The "Exists" symbol)

• Used for: Saying that something is real, even if it's just one.

• Meaning: "There is at least one of these out there."

• How to use: It almost always goes with "AND" (∧).

• Example: ∃x (King(x) ∧ Evil(x)) means "There exists at least one $x$ who is BOTH a King
AND Evil."

Que 8. Discuss the structured knowledge representation and its advantages

Ans.

What is it? Structured KR means organizing knowledge into neat packages, rather than just
having a giant list of random facts. It's like using a filing cabinet instead of dumping all your
papers on the floor.

Common types:

• Frames: Like an ID card for a concept (e.g., a "Car" frame has slots for color, model,
wheels).

• Semantic Nets: A diagram connecting ideas with arrows (e.g., a "Cat" circle connected
by an "is-a" arrow to an "Animal" circle).

Advantages:

1. Easier to understand: Humans can read and fix it easily because it's organized logically.

2. Inheritance (The best part): If you know a "Sparrow" is a type of "Bird," the Sparrow
automatically gets all the Bird info (wings, feathers, lays eggs) without you having to
write it all down again.

3. Faster: The AI knows exactly where to look for information because it's filed away
correctly.
Que 9. Differentiate between universal and existential quantifiers.

Ans.

Feature Universal Quantifier (∀) Existential Quantifier (∃)

Simple Name The "For All" symbol. The "There Exists" symbol.

At least 1 thing (maybe more, but at

Logical Partner Uses If-Then (→). Uses AND (∧).

Example "Everyone likes ice cream." "Someone likes spinach."

To prove it Find just ONE person who hates ice You have to prove NOBODY likes
WRONG cream. spinach.