Knowledge RepResentation
and Reasoning
Knowledge representation and reasoning form the
foundation of artificial intelligence (AI), allowing systems to
process, infer, and reason about knowledge to make decisions
or solve problems. This unit focuses on different methods and
techniques used to represent and reason about knowledge in
AI systems.
Knowledge Representation
Knowledge representation refers to encoding information
about the world in a form that computers can use to solve
complex tasks, such as diagnosing diseases, understanding
language, or playing chess. It must be structured, expressive,
and computationally efficient.
1. Propositional Logic
Propositional logic represents facts using propositions or
declarative statements that are either true or false.
• Key Features:
o Logical connectives: AND (), OR (), NOT (),
implication (), and equivalence ().
o Simple and easy to implement but lacks the ability
to represent relationships between objects.
• Example Use: Representing basic facts and their
relationships, such as “It is raining” or “If it rains, the
ground is wet.”
2. First-Order Predicate Logic (FOPL)
FOPL extends propositional logic by incorporating objects,
predicates, and quantifiers.
• Predicates: Represent relationships or properties, such
as .
• Quantifiers: Include “for all” () and “there exists” ().
FOPL is more expressive, allowing representation of
statements like “All humans are mortal.”
3. Semantic Networks
A semantic network represents knowledge as a graph where:
• Nodes denote concepts or entities.
• Edges represent relationships between nodes, such as
“IS-A” (hierarchy) or “PART-OF.”
• Applications:
o Natural language understanding
o Knowledge visualization
4. Frames
Frames are structured data forms for representing
stereotypical situations, akin to templates. They consist of:
• Slots: Attributes or properties of a concept.
• Values: Specific information for each slot.
• Advantages:
o Easy to represent default values and exceptions.
o Useful in object-oriented systems.
Comparison of Techniques
Technique Key Feature Strengths Limitations
Propositional True/false Lacks relational
Simplicity
Logic propositions representation
Objects,
First-Order Computationally
predicates, Expressive
Logic expensive
quantifiers
Semantic Graph Visual and Limited handling
Networks representation relational of uncertainty
Defaults
Structured
Frames and Limited flexibility
templates
exceptions
Reasoning Techniques
Reasoning refers to deriving conclusions or making decisions
based on represented knowledge. Different techniques
handle reasoning under certainty and uncertainty.
1. Logical Reasoning
Logical reasoning uses formal logic to draw conclusions from
known facts.
• Deductive Reasoning:
o Moves from general rules to specific conclusions.
o Example: If all humans are mortal, and Socrates is a
human, then Socrates is mortal.
• Inductive Reasoning:
o Generalizes from specific instances to broader
rules.
o Example: Observing multiple birds flying and
concluding that all birds can fly.
2. Probabilistic Reasoning
Probabilistic reasoning is used when certainty is unattainable,
allowing decisions based on likelihoods.
• Bayes' Theorem:
o Updates the probability of a hypothesis based on
new evidence.
o Formula: .
• Bayesian Networks:
o Graphical models representing probabilistic
dependencies among variables.
o Applications include medical diagnosis and risk
assessment.
3. Handling Uncertainty in AI
AI systems often operate in environments with incomplete or
ambiguous information. Techniques to handle uncertainty
include:
• Fuzzy Logic:
o Represents degrees of truth rather than binary
true/false values.
o Commonly used in control systems (e.g., washing
machines, air conditioners).
• Dempster-Shafer Theory:
o Combines evidence from different sources to
compute degrees of belief.
o Flexible in representing both uncertainty and
ignorance.
4. Comparison of Reasoning Techniques
Technique Key Feature Strengths Limitations
General to Requires
Deductive Certain
specific complete
Reasoning conclusions
inference knowledge
Specific to
Inductive Useful for Uncertain
general
Reasoning hypotheses conclusions
inference
Likelihood-
Probabilistic Handles Computational
based
Reasoning uncertainty complexity
reasoning
Flexible
Degrees of Requires careful
Fuzzy Logic decision-
truth calibration
making
Expert Systems
Expert systems are AI applications that mimic human experts
to solve specific domain problems. They combine knowledge
representation and reasoning techniques to provide decision
support.
Components of Expert Systems
1. Knowledge Base:
o Stores facts and rules about a domain.
2. Inference Engine:
o Applies logical reasoning to the knowledge base to
derive conclusions.
3. User Interface:
o Allows users to interact with the system.
Architecture of Expert Systems
• Rule-Based Systems:
o Use if-then rules for decision-making.
o Example: Medical diagnosis systems.
• Model-Based Systems:
o Use structural and behavioral models of the
domain.
o Example: Fault diagnosis in machinery.
Applications of Expert Systems
1. Medical Diagnosis:
o Identifying diseases based on symptoms.
2. Financial Analysis:
o Risk assessment and investment planning.
3. Engineering:
o Fault detection and process control.
