2 Marks Questions (2-3 lines, in points)

1. State any two ways how knowledge can be represented?

 Semantic Networks: Represent knowledge using nodes (concepts) and links

(relationships).

 Frames: Organize knowledge into structured units with slots for attributes and
values.

2. Describe the role of inheritance in engineering branch domain.

 Inheritance allows common properties of the engineering domain to be shared with

specific branches (e.g., Civil, Mechanical).

 This reduces redundancy and organizes knowledge hierarchically.

3. What is the role of constraints in knowledge representation?

 Constraints restrict the values or relationships among variables.

 They ensure that only valid and meaningful solutions are considered.

4. What type of constraints are used in Graph coloring problem?

 Adjacent nodes must not have the same color (binary constraint).

 Ensures all connected nodes are assigned different colors.

5. What are the inferences used under uncertainty? Explain with an example.

 Probabilistic inferences like Bayesian inference are used.

 Example: If it’s cloudy, the probability of rain increases based on past data.

6. State the need for conditional probability?

 Conditional probability updates beliefs when new evidence is available.

 It allows reasoning under uncertainty by quantifying event likelihoods given

conditions.

5 Marks Questions (8-10 lines, in points or tables)

1. What is the need for logic in knowledge representation? Explain.

 Logic provides a formal structure to encode facts and rules.

 Enables automated reasoning and deduction of new knowledge.

 Ensures consistency and detects contradictions in the knowledge base.

 Supports inference mechanisms like modus ponens.

 Allows for precise, unambiguous representation of relationships.

 Facilitates complex problem-solving and decision-making.

 Widely used logics: propositional and predicate logic.

 Helps maintain reliability and correctness of AI systems.

2. List the advantages and disadvantages of Semantic Networks.

Advantages Disadvantages

Intuitive visualization of Can become complex as the network

relationships grows

Easy inheritance of properties Ambiguity if link types are not clearly

defined

Flexible for adding new May lack expressive power for

information complex logic

Quick retrieval of related Inefficient for very large knowledge

concepts bases

3. What are the various types of links available in semantic network?

 IS-A: Represents class-subclass relationships (inheritance).

 HAS-A: Connects objects to their properties or parts.

 PART-OF: Indicates component relationships.

 Instance-of: Links an object to its class.

 Causal/Temporal: Represents cause-effect or time-based relationships.

4. How inheritance is represented in Semantic Net? Explain the same with an

example.

 Inheritance is shown using IS-A links (subclass to superclass).

  Properties defined at the superclass are inherited by subclasses.

 Example:

o "Dog IS-A Animal" (Dog inherits all Animal properties).

o If "Animal" has "can move", "Dog" also "can move" through inheritance.

 Reduces redundancy and simplifies knowledge updates.

5. How knowledge representation can be done using Frames? Explain.

 Frames represent knowledge as structured units with slots for attributes.

 Each frame corresponds to an object or concept.

 Slots can contain values or point to other frames.

 Supports inheritance: subclasses inherit slots from parent frames.

 Example: "Car" frame with slots for "Color", "Engine Type", "Number of Wheels".

 "Sports Car" frame inherits from "Car" and adds specific slots.

 Organizes knowledge in a modular and reusable way.

6. Write short notes on Baye's probabilistic inferences.

 Bayes' inference updates the probability of a hypothesis given new evidence.

 Uses Bayes' theorem to combine prior knowledge and observed data.

 Calculates posterior probability for decision-making.

 Handles uncertainty in AI systems (e.g., medical diagnosis).

 Allows explicit belief updating as more evidence is gathered.

10 Marks Questions (15-20 lines, in points or tables)

1. What is the need for knowledge representation? What are the various issues
associated with it?

Need for Knowledge Representation:

 Enables machines to store and organize information about the world.

 Supports reasoning, learning, and problem-solving.

  Facilitates efficient retrieval and manipulation of knowledge.

 Essential for context understanding and decision-making in AI.

 Bridges the gap between raw data and actionable intelligence.

Issues in Knowledge Representation:

Issue Description

Expressivene Capturing all necessary details without excessive

ss complexity.

Scalability Managing and searching large knowledge bases

efficiently.

Ambiguity Handling vagueness and multiple meanings in concepts

or language.

Incompletene Representing and reasoning with missing or partial

ss information.

Uncertainty Dealing with probabilistic or uncertain knowledge.

Consistency Avoiding contradictions within the knowledge base.

Interoperabili Ensuring different representations can work together.

ty

Maintainabilit Keeping knowledge up to date as the world changes.

y

2. Create a semantic network to represent animal classifications to represent

different animals like "Mammal", "Bird", "Fish" and their properties like "has
fur", "lays eggs", etc.

 Nodes: Animal, Mammal, Bird, Fish, Dog, Parrot, Shark.

 IS-A Links:

o Mammal IS-A Animal

o Bird IS-A Animal

o Fish IS-A Animal

o Dog IS-A Mammal

o Parrot IS-A Bird

 o Shark IS-A Fish

 HAS-A Links:

o Mammal HAS-A "has fur"

o Bird HAS-A "lays eggs", "can fly"

o Fish HAS-A "has scales", "can swim"

 Inheritance:

o Dog inherits "has fur"

o Parrot inherits "lays eggs" and "can fly"

o Shark inherits "has scales" and "can swim"

 Diagram Description:

o "Animal" at the root, branching to "Mammal", "Bird", "Fish", then to specific

animals, with property links attached.

3. For the following knowledge, construct semantic network.

Statement Semantic Network Representation

Humans, animals, birds are living things IS-A links from Human, Animal, Bird to Living Thing; Living
who can breathe, eat Thing HAS-A "can breathe", "can eat"

All birds can fly Bird HAS-A "can fly"

Man & woman are human who have 2 Man IS-A Human, Woman IS-A Human; Human HAS-A "2 legs"
legs

Cat has fur & is an animal Cat IS-A Animal; Cat HAS-A "fur"

All animals have skin & can move Animal HAS-A "skin", "can move"

Giraffe is an animal, has long legs & is Giraffe IS-A Animal; Giraffe HAS-A "long legs", "is tall"
tall

Parrot is a bird & is green in color Parrot IS-A Bird; Parrot HAS-A "green color"

 Inheritance: Properties like "can breathe", "can eat", "can move" are inherited by
subclasses.

 Structure: Central "Living Thing" node, branching to Human, Animal, Bird, then to
Man, Woman, Cat, Giraffe, Parrot with property links.
4. Explain how reasoning is accomplished using semantic networks.

 Reasoning is performed by traversing IS-A and HAS-A links in the network.

 Inheritance allows subclasses to acquire properties from superclasses.

 To answer queries, the system follows links to infer relationships or attributes.

 Example: "Dog IS-A Mammal", "Mammal HAS-A warm blood" → "Dog HAS-A warm
blood".

 Supports property inheritance, forward and backward chaining.

 Can resolve queries like "Which animals can fly?" by following links from "Bird".

 Handles exceptions by adding specific links that override inherited properties.

 Enables efficient deduction and supports knowledge-based reasoning in AI.

5. Explain how a semantic network can be used to represent and query

information about different types of vehicles, including their properties like
"has wheels," "can fly," and their relationship to categories like "Land Vehicle,"
"Air Vehicle," and "Water Vehicle." Provide an example diagram.

 Nodes: Vehicle, Land Vehicle, Air Vehicle, Water Vehicle, Car, Airplane, Boat.

 IS-A Links:

o Land Vehicle IS-A Vehicle

o Air Vehicle IS-A Vehicle

o Water Vehicle IS-A Vehicle

o Car IS-A Land Vehicle

o Airplane IS-A Air Vehicle

o Boat IS-A Water Vehicle

 HAS-A Links:

o Land Vehicle HAS-A "has wheels"

o Air Vehicle HAS-A "can fly"

o Water Vehicle HAS-A "can float"

o Car HAS-A "engine type"

o Boat HAS-A "propeller"

  Query Example: To find vehicles that can fly, follow IS-A from "Vehicle" to "Air
Vehicle", and retrieve all subclasses (e.g., Airplane).

 Diagram Description: "Vehicle" at the top, branching to Land, Air, Water Vehicles,
then to specific types, with property links.

6. Implement a basic Dempster-Shafer theory example by combining evidence

from multiple sources to determine the likelihood of an event and use two
pieces of evidence to update the belief in a hypothesis.

 Scenario: Will it rain?

 Evidence:

o Sensor A: Belief in "rain" = 0.6, "no rain" = 0.2, "uncertain" = 0.2

o Sensor B: Belief in "rain" = 0.7, "no rain" = 0.1, "uncertain" = 0.2

 Combination (Dempster’s Rule):

o Combine beliefs for "rain": (0.6 × 0.7) + (0.6 × 0.2) + (0.2 × 0.7) = 0.42 +
0.12 + 0.14 = 0.68 (before normalization)

o Normalize to ensure total probability is 1.

 Result:

o Combined belief in "rain" increases, uncertainty decreases.

o System can now make a stronger prediction based on aggregated evidence.

 Advantage: Handles uncertainty and partial knowledge, giving a more robust

estimate than single-source evidence.

 Application: Used in expert systems where evidence may be incomplete or

conflicting.

Let me know if you need further formatting or want any answers expanded or shortened!