trí tuệ nhân tạo cao hoàng trứ chương ter4 knowledge representation sinhvienzone com

om

nh
Vi
en

Zo

ne

.C

Knowledge Representation
Si

Chapter 4

SinhVienZone.com

/>

om

What is KR?

ne

.C

R. Davis, H. Schrobe, P. Szolovits (1993):

Zo

1. A surrogate

nh
Vi
en

2. A set of ontological commitments
3. A fragmentary theory of intelligent reasoning
4. A medium for efficient computation

Si

5. A medium of human expressions

2

SinhVienZone.com

/>

om

Representation and Mapping

ne

.C

• Facts: things we want to represent.

Si

nh
Vi
en

Zo

• Representations of facts: things we can manipulate.

3

SinhVienZone.com

/>

Internal
Representations

Zo

Facts

ne

.C

om

Representation and Mapping

Si

nh
Vi
en

English
understanding

reasoning
programs

English
generation

English
Representations

4

SinhVienZone.com

/>

om

Representation and Mapping

.C

desired real reasoning

Zo

ne

Initial
facts

nh
Vi
en

forward
representation
mapping

Si

Internal
representations
of initial facts

Final
facts

backward

representation
mapping

operation
of program

Internal
representations
of final facts

5

SinhVienZone.com

/>

om

Representation and Mapping

ne
Zo

Si

nh
Vi
en

• Every dog has a tail

.C

• Spot is a dog

Spot has a tail
6

SinhVienZone.com

/>

om

Representation and Mapping

.C

• Spot is a dog

Zo

• Every dog has a tail

ne

dog(Spot)

Si

nh
Vi
en

∀x: dog(x) → hastail(x)

hastail(Spot)

Spot has a tail
7

SinhVienZone.com

/>

om

Representation and Mapping

ne

.C

• Fact-representation mapping is not one-to-one.
• Good representation can make a reasoning program

Si

nh
Vi

en

Zo

trivial.

8

SinhVienZone.com

/>

om

Representation and Mapping

.C

The Multilated Checkerboard Problem

Zo

ne

“Consider a normal checker board from which two
squares, in opposite corners, have been removed.

Si

nh

Vi
en

The task is to cover all the remaining squares exactly
with donimoes, each of which covers two squares. No
overlapping, either of dominoes on top of each other or
of dominoes over the boundary of the multilated board
are allowed.
Can this task be done?”

9

SinhVienZone.com

/>

No. black squares
= 30
No. white square
= 32

Si

nh
Vi
en

Zo

ne

.C

om

Representation and Mapping

10

SinhVienZone.com

/>

Good representation:

nh
Vi
en

Inferential efficiency

Zo

Inferential adequacy

ne

Representational adequacy

Acquisitional efficiency

Si

•
•
•
•

.C

om

Representation and Mapping

11

SinhVienZone.com

/>

.C

Simple relational knowledge:

om

Approaches to KR

Si

nh
Vi
en

Zo

ne

• Provides very weak inferential capabilities.
• May serve as the input to powerful inference engines.

12

SinhVienZone.com

/>

om

Approaches to KR

.C

Inheritable knowledge:

ne

• Objects are organized into classes and classes are

Zo

organized in a generalization hierarchy.

Si

adequate.

nh
Vi
en

• Inheritance is a powerful form of inference, but not

13

SinhVienZone.com

/>

om

Approaches to KR

.C

Inferential knowledge:

ne

• Facts represented in a logical form, which facilitates

Zo

reasoning.

Si

nh
Vi
en

• An inference engine is required.

14

SinhVienZone.com

/>

om

Approaches to KR

.C

Procedural knowledge:

ne

• Representation of “how to make it” rather than “what

Zo

it is”.

nh
Vi
en

• May have inferential efficiency, but no inferential

Si

adequacy and acquisitional efficiency.

15

SinhVienZone.com

/>

om

Approaches to KR

.C

Choosing the Granularity:

Si

nh
Vi
en

Zo

ne

• High-level facts may not be adequate for inference.
• Low-level primitives may require a lot of storage.

16

SinhVienZone.com

/>

om

Homework

.C

Reading

Si

nh
Vi

en

Zo

ne

R. Davis, H. Schrobe, P. Szolovits (1993): “What is a knowledge
representation?”

17

SinhVienZone.com

/>