Artificial Intelligence - UNIT 3-Topic 1-Representation of knowledge

 

UNIT - III

 Representation of Knowledge

---

Part A: Introduction

---

✅ What is Knowledge Representation (KR)?

In Artificial Intelligence, Knowledge Representation (KR) refers to the method used to organize, store, and use knowledge in such a way that a machine or AI agent can understand, reason, and make decisions like a human.

---

✅ Why is KR Important?

• Helps AI systems understand the world and facts.
• Enables machines to perform reasoning, problem-solving, and learning.
• Converts real-world information into a structured format that computers can process.

---

Part B: Forms of Knowledge in AI

Type of Knowledge	Description	Example
Factual Knowledge	Basic facts or truths	“The sun rises in the east”
Procedural Knowledge	How to perform tasks or steps	“How to drive a car”
Heuristic Knowledge	Rules-of-thumb, educated guesses	“If traffic is heavy, leave early”
Meta-Knowledge	Knowledge about knowledge	“We know that humans learn from errors”
Declarative Knowledge	Describes things using statements	“John is a teacher”

---

Part C: Desirable Properties of Knowledge Representation

Property	Description
Representational Accuracy	Can represent all types of required knowledge correctly
Inferential Adequacy	Supports new knowledge inference using existing information
Inferential Efficiency	Allows efficient reasoning and searching
Acquisition Efficiency	Easy to input and update new knowledge

---

Part D: Major Techniques for Knowledge Representation

---

✅ 1. Logical Representation (Predicate Logic)

• Uses statements and symbols to represent facts.
• Good for formal reasoning.
• Example:
  Father(John, David) → "John is the father of David"

---

✅ 2. Semantic Networks

• Graph-based structures using nodes (objects/concepts) and links (relationships).
• Example:
  "Bird → is-a → Animal"
  "Bird → has → Wings"

---

✅ 3. Frames

• Data structures for representing stereotypical knowledge.
• Like an object with slots and values.
• Example:
  Frame: CAR
• Type: Vehicle
• Wheels: 4
• Fuel: Petrol/Diesel

---

4. Production Rules (If-Then Rules)

• Knowledge is represented as rules:

IF condition THEN action

• Example:
  IF temperature < 20 THEN wear a jacket

---

5. Constraint-Based Representation

• Uses constraints or conditions to restrict possible solutions.
• Useful in scheduling and puzzle-solving.
• Example:
  “Task A must be done before Task B”

---

6. Ontology-Based Representation

• Organizes knowledge using a hierarchy of concepts and relationships.
• Example: In medicine: Disease → Type → Symptoms → Treatment

---

Part E: Challenges in Knowledge Representation

Challenge	Explanation
Uncertainty	Not all knowledge is 100% true or complete
Ambiguity	Words or situations can have multiple meanings
Incompleteness	Some knowledge may be missing or not yet known
Dynamic Nature of World	Knowledge constantly changes; the system must adapt

---

Real-Life Applications of KR

Domain	How KR is Used
Expert Systems	Diagnosing diseases, troubleshooting technical problems
Robotics	Understanding objects and actions in the environment
Natural Language Processing (NLP)	Interpreting meaning of human language
Intelligent Agents	Making decisions based on rules, conditions, and knowledge

---

Summary

• Knowledge Representation is a key part of AI that helps machines store and use information like humans.
• Several techniques like logic, semantic nets, frames, and rules are used.
• A good KR system should be accurate, efficient, and easy to update.
• It is widely used in areas like expert systems, NLP, and robotics.

“Knowledge is power — and in AI, the way we represent that knowledge is how we give power to machines.”