37
TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN Topic 1: Introduction By : Shereena Arif Room : T2/8, Blok H Email : [email protected] / [email protected]

TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

  • Upload
    lee

  • View
    37

  • Download
    0

Embed Size (px)

DESCRIPTION

TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN. By : Shereena Arif Room : T2/8, Blok H Email : [email protected] / [email protected]. Topic 1: Introduction. Learning Outcomes. At the end of this topic, students will: Acquire good background understanding of concepts relating to KR - PowerPoint PPT Presentation

Citation preview

Page 1: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Topic 1: Introduction

By : Shereena ArifRoom : T2/8, Blok HEmail : [email protected] / [email protected]

Page 2: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Learning Outcomes

TP2623, [email protected]

2

At the end of this topic, students will: Acquire good background understanding

of concepts relating to KR Understand the concepts, definitions and

principles of KR

04/05/2011

Page 3: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

What is knowledge-based system (KBS)?

TP2623, [email protected]

3

KBS is a knowledge intensive computer program that contains aspects of human knowledge and expertise to perform tasks ordinarily done by human expert.

Emerged from the artificial intelligence (AI) domain.

04/05/2011

Page 4: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

What is Artificial Intelligence (AI)? AI is a field which

deals with the design of a program that makes it possible to perceive, reason and act.

Machine Learning

ImageProcessing

Speech Recognition

Etc.

Robotics

Natural Language

Understanding

Expert system

Knowledge-based system

AI

TP2623, [email protected]

4

04/05/2011

Page 5: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

What is knowledge?

TP2623, [email protected]

5

In 1970s, it was accepted that to make machines solve intellectual problems – had to know the solutions.

A machine has to have ‘knowledge’ i.e. ‘know how’ in some specific domain.

Knowledge is a theoretical or practical understanding of a subject or a domain.

04/05/2011

Page 6: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

What is knowledge?

TP2623, [email protected]

6

Concepts of the same category: the meaning of a sentence a belief a knowledge an information

They are about a property in some domain something that can be true or false in the world Ex. All humans are male or female Ex. Today’s lecture takes place in BK5

04/05/2011

Page 7: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

What is knowledge?

TP2623, [email protected]

7

The meaning of a descriptive natural language (NL) sentence is an information Ex. The world is flat Ex. The world is a sphere

Not all sentences have a property as meaning: Ex. The truck grew white smell Ex. Are there sentences that have meaning and

do not describe a property of the world? Ex. Search two of them on this slide!

Questions, commands, ..04/05/2011

Page 8: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

What is knowledge?

TP2623, [email protected]

8

Information are omnipresent in human minds and reasoning

Humans store and process massive amounts of information

Humans maintain a knowledge base of information which they consider to be true in the actual world

But humans have many other stores of “information”

04/05/2011

Page 9: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

What is knowledge?

TP2623, [email protected]

9

Humans store other “informations”, not considered to be true in the world: a knowledge : Ex. The world is a sphere a belief: Ex. Osama Laden is now dead a goal: Ex. I will be rich in 2012 a hope: Ex. There will be peace in Palestine in

2013 beliefs about beliefs: Ex. In the middle ages, all

people believed the earth was flat promises, obligations, . . .

These are called propositional attitudes04/05/2011

Page 10: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Why do we need knowledge?

TP2623, [email protected]

10

Humans compute information all the time: e.g. whenever interpreting speech or text

Humans evaluate truth of information: Ex. Next lecture is about monkey and takes

place in the National Zoo Humans update their knowledge base

with information accordingly

04/05/2011

Page 11: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Why do we need knowledge?

TP2623, [email protected]

11

Humans can reason with information to solve problems and tasks Ex. From “all KR lectures take place in room BK5”,

infer “the lecture of next week takes place in BK5” whether information are believed or not: Ex. If Bin Laden is still alive, there will be another

suicide bombing, this time in Paris. Humans have goals and they use knowledge

to compute plans of actions to reach goals Ex. To attend the KR lecture, you came to BK5

located at Block B. Ex. To become rich, I will create spin-off company

04/05/2011

Page 12: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Other sorts of knowledge

TP2623, [email protected]

12

Defeasible knowledge Ex. Birds can fly.

Probabilistic knowledge Ex. Chances are high that tomorrow will be

sunny. Vague knowledge

Ex. The weather is beautiful. The house too. Declarative versus Procedural

Knowledge Important issues in AI and psychology

04/05/2011

Page 13: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Types of Knowledge

TP2623, [email protected]

13

Declarative Procedural Semantic Episodic Commonsense Heuristic

04/05/2011

Page 14: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Declarative Knowledge

TP2623, [email protected]

14

Surface level knowledge Attributes and characteristics Example:

describe this room

04/05/2011

Page 15: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Procedural Knowledge

TP2623, [email protected]

15

Processes and Procedures Skills Example:

How did you get to this room How do you drive a car

04/05/2011

Page 16: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Semantic Knowledge

TP2623, [email protected]

16

Relations and connections How things fit together Example:

What do you think about when I say “room” and “door”?

The relationship between signs, symbols and what they represent

04/05/2011

Page 17: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Episodic Knowledge

TP2623, [email protected]

17

Stories, cases, examples Based on experiences Example:

Tell me about your birthday celebration

04/05/2011

Page 18: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Commonsense Knowledge

TP2623, [email protected]

18

General knowledge about the world Obvious to most people Built up over time Example:

Buildings contain rooms Rooms are smaller than buildings

04/05/2011

Page 19: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Heuristic Knowledge

TP2623, [email protected]

19

Rule-of-thumb Directional accuracy; doesn’t guarantee

outcome IF-THEN rules Example:

If the lights are out, no one is home.

04/05/2011

Page 20: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Knowledge in KBSs

Conventional Programming

Knowledge-Based Systems

20

TP2623, [email protected]

Algorithms

+ Data Structures

= Programs

Knowledge

+ Inference

= KBS

04/05/2011

Page 21: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Declarative & Procedural K.

TP2623, [email protected]

21

Declarative info: properties about world; “declarative” means “descriptive” The "what" or content of learning; knowing a piece of

information - a fact, a concept, or a label. Humans have lots of procedural knowledge

We remember certain procedures that yield useful goals When the goal rises, we execute the procedure (without

reasoning) Easy and Fast!!

Old controversy II: Which knowledge is most important for AI? Declarative AI-languages (e.g. logics) versus procedural AI-languages (e.g. production systems).

04/05/2011

Page 22: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Procedural Knowledge

TP2623, [email protected]

22

Ex. Goal: clean clothes; Your laundry procedure

Ex. Goal: to eat; your restaurant procedure, your menu procedure, . . .

Ex. Goal: to have information; your text or speech interpretation procedure

Ex. Calculation: 22 x 45 = 22 + 22 +……+22 (45 times) Our procedure 2 2

x 4 5 1 1 0 8 8 9 9 0

04/05/2011

Page 23: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Procedural Knowledge

TP2623, [email protected]

23

What is a procedure? A known pattern of actions

Subconscious processing: Sometimes we cannot control executions of

a procedure Ex. Count the number of words in the

following sentence without interpreting the sentence:“I said do not interpret this sentence!”

04/05/2011

Page 24: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

What is knowledge representation?

TP2623, [email protected]

24

Many definitions…. KR is the study of:

How knowledge about the world can be represented, &

What kinds of reasoning can be done with that knowledge

It is the process of focusing and representing only the essential part of real world problem into some representation.

04/05/2011

Page 25: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

What is reasoning?

TP2623, [email protected]

25

A method of thinking Process of constructing new configurations

(sentences) from old ones proper reasoning ensures that the new

configurations represent facts that actually follow from the facts that the old configurations represent

this relationship is called entailment and can be expressed asKB |= alpha knowledge base KB entails the sentence alpha

04/05/2011

Page 26: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

What is knowledge engineering?

TP2623, [email protected]

26

The disciplines for developing/building KBSs.

The activities – transferring & transforming aspects of problem solving expertise from knowledge sources into a program or system.

A number of methodologies to support knowledge engineering activities have been proposed.

04/05/2011

Page 27: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Principles of KR

TP2623, [email protected]

27

Five basic principles about KR & their role in artificial intelligence:

1. A KR is a surrogate2. A KR is a set of ontological commitment3. A KR is a fragmentary theory of

intelligence reasoning4. A KR is a medium for efficient

computation5. A KR is a medium of human expression

04/05/2011

Page 28: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

KR is a surrogate

TP2623, [email protected]

28

Physical objects, events and relationships which cannot be stored directly in a computer – represented by symbols – serve as a surrogates for the external things.

04/05/2011

Page 29: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Ontological commitments

TP2623, [email protected]

29

Ontology is the study of existence For a database of knowledge base,

ontology determines the things that exist in the application domain

Those categories represents the ontological commitment of the designer

04/05/2011

Page 30: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Fragmentary theory of intelligence reasoning

TP2623, [email protected]

30

To support reasoning about the things in a domain, a KR must describe their behavior and interaction.

04/05/2011

Page 31: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Medium for efficient computation

TP2623, [email protected]

31

Besides representing knowledge, an AI system must encode knowledge in a form that can be processed efficiently

04/05/2011

Page 32: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Medium for human expression

TP2623, [email protected]

32

A good KR language should facilitate communication between the knowledge engineer who understand AI and the domain experts who understand the application

04/05/2011

Page 33: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

KR Technique

TP2623, [email protected]

33

A number of early knowledge representation techniques are: Rules Semantic net Frames Logic

04/05/2011

Page 34: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

What will you learn next…..

Logic (propositional + description)

Semantic Net Frame Rules Knowledge

engineering

Ontology and ontology engineering

Knowledge representation for the web – Semantic Web

Flex

34

TP2623, [email protected] 04/05/2011

Page 35: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

Pendekatan

TP2623, [email protected]

35

Pengendalian Kursus Kuliah & Makmal

Penilaian Tugasan Makmal (30%) Kehadiran (10%) Peperiksaan akhir (60%)

04/05/2011

Page 36: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

References

TP2623, [email protected]

36

Brachman, R.J. & Levesque, H. J. 2004. Knowledge Representation & Reasoning. New York: Morgan Kaufmann.

Russell, S. & Norvig, P. 2003. Artificial Intelligence: A Modern Approach. New Jersey: Prentice Hall.

Antoniou, G. & van Harmelen, F. 2004. A Semantic Web Primer. Cambridge: MIT Press

Negnevistky, M. 2002. Artificial Intelligence. Harlow: Addison-Wesley.

Sowa, J. F. and Dietz D. 1999. Knowledge Representation: Logical, Philosophical, and Computational Foundations. New York: Broke/Cole Pub.

Lain-lain – artikel dalam jurnal dan sumber daripada Internet

04/05/2011

Page 37: TP 2623 : PERWAKILAN & PENAAKULAN PENGETAHUAN

END OF TOPIC 1

THANK YOU..