Scribd
Upload
98 views

Assumption of (Traditional) AI Work Is That

The document discusses knowledge representation in artificial intelligence. It covers several key points: 1) Traditional AI assumes knowledge can be represented through symbolic structures like data structures representing objects, concepts, facts, and rules. Intelligent behavior is achieved by manipulating these symbols. 2) Representation is an important issue in AI as choosing the wrong representation could cause a project to fail. Common representation schemes include logic, frames, semantic networks, and rules. 3) For knowledge representation to support problem solving, the representation needs to have adequate expressiveness to represent the required knowledge, allow inferences to be made, and efficiently acquire and manipulate knowledge.

Uploaded by

Nitesh Mishra
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
98 views

Assumption of (Traditional) AI Work Is That

The document discusses knowledge representation in artificial intelligence. It covers several key points: 1) Traditional AI assumes knowledge can be represented through symbolic structures like data structures representing objects, concepts, facts, and rules. Intelligent behavior is achieved by manipulating these symbols. 2) Representation is an important issue in AI as choosing the wrong representation could cause a project to fail. Common representation schemes include logic, frames, semantic networks, and rules. 3) For knowledge representation to support problem solving, the representation needs to have adequate expressiveness to represent the required knowledge, allow inferences to be made, and efficiently acquire and manipulate knowledge.

Uploaded by

Nitesh Mishra
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 20

Introduction

Assumption of (traditional) AI work is that:


Knowledge may be represented as “symbol
structures” (essentially, complex data structures)
representing bits of knowledge (objects, concepts,
facts, rules, strategies..).
E.g., “red” represents colour red.
“car1” represents my car.
red(car1) represents fact that my car is red.
Intelligent behavior can be achieved through
manipulation of symbol structures
Representation Representation Representation

Think about knowledge, rather than data in AI


Facts
Procedures
Meaning
Cannot have intelligence without knowledge
Always been very important in AI
Choosing the wrong representation
Could lead to a project failing
Still a lot of work done on representation issues
Representations for
Problem solving techniques
For certain problem solving techniques
 The “best” representation has already been worked out
 Often it is an obvious requirement of the technique
 Or a requirement of the programming language (e.g., Prolog)
Examples:
 First order theorem proving (first order logic)
 Inductive logic programming (logic programs)
 Neural networks learning (neural networks)
But what if you have a new project?
 What kind of general representations schemes are there?

Representational adequacy
Consider the following facts:
John believes no-one likes brussel sprouts.
Most children believe in Santa..
Can all be represented as a string! But hard then to
manipulate and draw conclusions.
How do we represent these formally in a way that can
be manipulated in a computer program?
The ability to represent all kinds of knowledge that are
needed in that domain.
Well-defined

syntax/semantics
Knowledge representation languages should have
precise syntax and semantics.
You must know exactly what an expression means
in terms of objects in the real world.

Real World Real World

Map to KR language Map back to real world

New
Representation Inference conclusions
of facts in World
Computer
Computer
Well defined syntax/semantics
Suppose we have decided that “red1” refers to a dark
red colour, “car1” is my car, car2 is another..
Syntax of language will tell you which of following is
legal: red1(car1), red1 car1, car1(red1),
red1(car1 & car2)?
Semantics of language tells you exactly what an
expression means - e.g., Pred(Arg) means that the
property refered to by Pred applies to the object
refered to by Arg. E.g., property “dark red” applies
to my car.
Inferential Adequacy
Representing knowledge not very interesting unless you can
use it to make inferences:
 Draw new conclusions from existing facts.
 “If its raining John never goes out” + “It’s raining today” so..
 Come up with solutions to complex problems, using the
represented knowledge.
Inferential adequacy refers to how easy it is to draw
inferences using represented knowledge.
Representing everything as natural language strings has
good representational adequacy and naturalness, but very
poor inferential adequacy.
The ability to manipulate the representational structures in
such a way to derive new structures corresponding to new
knowledge inferred from old
Inferential Efficiency
You may be able, in principle, to make complex
deductions given knowledge represented in a
sophisticated language.
But it may be just too inefficient.
Generally the more complex the possible deductions,
the less efficient will be the reasoned.
Need representation and inference system sufficient
for the task, without being hopelessly inefficient.
The ability to incorporate into knowledge structure
new information that help the inference mechanism
in most promising direction
Acquisitioned efficiency
The ability to acquire new information easily
The new knowledge can be easily added to database
The single system not optimize all the capabilities for
all kind of knowledge has not be found as a result
multiple techniques for knowledge representations
exists.
Knowledge acquisition

knowledge engineering: building expert systems


knowledge acquisition: process of extracting knowledge
from an expert, organizing it, and encoding it into a
knowledge base
knowledge elicitation: extracting knowledge from an expert
knowledge acquisition is the principle bottleneck in expert
system development
many techniques and theories about how to best do this
more tools are appearing to help in this
 early example: inductive inference tables
acquisition
Expert
data, problems,
questions

knowledge
concepts Formalized
solutions structured
knowledge
Knowledge KNOWLEDGE
engineer BASE

Needs,
usability,
feedback

Prototypes,
needs queries

End user
Also: other experts, literature
acquisition

1. Collect: (elicitation)
- getting the knowledge out of the expert
- most difficult step
- lots of strategies
2. Interpret:
- review collected knowledge, organize, filter
3. Analyze:
- determining types of knowledge, conceptual relationships
- determining appropriate knowledge represention & inference
structure
4. Design:
- extracting more knowledge after using above principles

Lets look at these in more detail...


Main KR Approaches
Logic
Frames/Semantic Netorks/Objects
Rule-based systems
Logic as a Knowledge
Representation Language
A Logic is a formal language, with precisely defined syntax and
semantics, which supports sound inference. Independent of
domain of application.
Different logics exist, which allow you to represent different
kinds of things, and which allow more or less efficient
inference.
 propositional logic, predicate logic, temporal logic, modal logic,
description logic..
But representing some things in logic may not be very natural,
and inferences may not be efficient. More specialised languages
may be better..
Propositional logic
In general a logic is defined by
syntax: what expressions are allowed in the language.
Semantics: what they mean, in terms of a mapping to
real world
proof theory: how we can draw new conclusions from
existing statements in the logic.
Propositional logic is the simplest..
Propositional Logic: Syntax
Symbols (e.g., letters, words) are used to represent facts
about the world, e.g.,
 “P” represents the fact “Andrew likes chocolate”
 “Q” represents the fact “Andrew has chocolate”
These are called atomic propositions
Logical connectives are used to represent and: , or:  , if-
then: , not: .
Statements or sentences in the language are constructed
from atomic propositions and logical connectives.
 P  Q “Andrew likes chocolate and he doesn’t have any.”
 P Q “If Andrew likes chocolate then Andrew has chocolate”
Propositional Logic: Semantics
What does it all mean?
Sentences in propositional logic tell you about what
is true or false.
P  Q means that both P and Q are true.
P  Q means that either P or Q is true (or both)
P  Q means that if P is true, so is Q.
This is all formally defined using truth tables.

We now know exactly what is meant in


XY XvY terms of the truth of the elementary
TT T propositions when we get a sentence in
TF T the language (e.g., P => Q v R).
FT T
FF F
Proof Theory
How do we draw new conclusions from existing
supplied facts?
We can define inference rules, which are guaranteed
to give true conclusions given true premises.
For propositional logic useful one is modus ponens:

If A is true and A=> B is true, B


A, A then conclude B is
—————————

true. B
Summary
Intelligent systems require that we have
Knowledge formally represented
New inferences/conclusions possible.
Formal languages have been developed to support
knowledge representation.
One important one is the use of logic - very general
purpose way to formally represent truths about the
world, and draw sound conclusions from these.
ISSUES in KR
Are any attributes of objects that they are occur
almost in every problem domain.
Are there any important relationship that exits
among attributes of objects.
At what level the knowledge to be represented
How the set of objects to be represented
How the relevant data is accessed from the database
when needed.

You might also like