Вы находитесь на странице: 1из 44

Expert System

Terminology

Knowledge Engineering The discipline of


acquiring, encoding and using human domain
knowledge to develop a computer application
Expert System A computer program that uses
domain knowledge to perform a specific task
usually human experts perform
Knowledge Base a set of rules and facts
describing the domain of an application
Inference engine a program that imposes a
general control strategy on how the system is
working
Working memory a set of facts describing a
particular consultation

WHAT ARE
EXPERT SYSTEMS?
An Expert System (ES) is a computerbased system (mainly software) that uses
knowledge and facts, and apply an
appropriate reasoning technique
(inferencing) to solve problems in a given
field (domain) that normally require the
services of human experts.
3

What is Expert System


Contd

A computer application that performs a task that


would otherwise be performed by a human expert.

A model and associated procedure that exhibits,


within a specific domain, a degree of expertise in
problem solving that is comparable to that of a
human expert. Ignizio

An expert system is a computer system which


emulates the decision-making ability of a human
expert.-Giarratono
4

So, what IS ES

A system which employs human expertise


captured in a CBIS to solve problems
which usually require human expertise.
An expert system either supports or
automates decision making in an area of
which experts perform better than non
experts. It is also known as "Expert
Computing Systems", or "Knowledge
Based Systems".
5

In other words..ES

are computer programs had try to replicate


knowledge and skills of humans expert in some
area, and then attempt to solve problems in this
area (the way human expert to do)
Take root from cognitive science the study of
human mind using a combination of AI and
psychology
Today, ES is the most successful application of AI
real world problem solving problems in medicine,
chemistry, finance and even in space.
Allows company save their cost per year.

BRIEF HISTORY
Examples of the early and famous expert
systems
DENDRAL
- Stanford Univ. (1965)

CADACEUS - Univ. of Pittsburgh (1970)

Analysis of chemical compunds


Rule-based system

Diagnosis of human internal diseases

MYCYSMA

- MIT (1971)

Symbolic mathematical analysis

Brief History (Contd)

MYCIN

Diagnosis and treatment of infectious blood diseases


Rule-based system

EMYCIN

- Stanford Univ. (1978)

Evolved from (Empty) Mycin


First ES Shell

XCON

- Stanford Univ. (1972)

- DEC (1980)

configures VAX computer system


First ES used commercially

1943 Post, E.L proved that any computable


problem can be solved using simple IF-THEN rules

1962 General Problem Solver (GPS) by A.Newell


and H.Simon

Computer Vs Mind

The whole is more than a sum of its parts

Computer use pre-programmed algorithms


Obviously, our mind do not work like computers:
different people usually solve problems in their
own ways, and you even may solve the same
problem differently every time
Even if we begin to know how neural networks
work, it is hard to scale up and understand how
they form thoughts
Some scientists believe that cognition (or
intelligence) is somewhat independent of the
architectural implementation (i.e. not necessarily
should use neural networks)
9

A Human Experts

Consider several examples:

A doctor
Chess grands-master
Financial wizard
A chief

What is different between them? Can you say


that one is more intelligent that another?

10

Intelligent Vs Expertise

Expertise and intelligence are not the


same things (although they are related)
Expertise requires long time to learn
Expertise is a large amount of knowledge
(in some domain)
Expertise is easily recalled
Intelligence allows you to use your
expertise (apply the knowledge)
Expertise enables you to find solution
much faster
11

Data -> Information -> Knowledge

DATA- measurements or records


about events. Can be numerical,
alphabetical, images, sounds, etc.
INFORMATION analyzed and
organized data such that we know its
characteristics
KNOWLEDGE information put into a
specific context
12

Levels of Abstraction

High
knowledge
Abstraction

information
low

Data

Data, Information and


Knowledge can be
classified by levels of
abstraction and quantity
Knowledge is the most
abstract and exists in the
smallest quantity
Knowledge itself can have
levels of abstraction :
concrete (knowledge about
the specific problem),
domain specific (class of
problems) and abstract
(many classes of problems)
13

Categorized of Knowledge

Symbolic, sub-symbolic & hybrid

Symbolic represented as discrete unit


(explicit)-> expressed in formal
language
Sub-symbolic cannot be represented
by single symbols (implicit), -> stored in
a form of information
Hybrid -> systems involved both types
of knowledge representation
14

Categorized of Knowledge (cont..)

Declarative & Procedural


Declarative -> Knowledge are facts (objects, events)
that we observe in the environment. Can describe
current state of a problem
e.g. : I' am a human or the sun shining
Procedural -> knowledge represents the skills or rules
we use to operate with and reason using facts
e.g. : IF I' am a human THEN I have two legs
Apparently, there are separate areas in the human brain
for storing declarative and procedural knowledge
Rules (procedural) are usually harder to describe

15

Basic Concept of ES

Expertise

Experts

Ability to reason-> component called the inference engine

Rules

Involves 4 activities -> knowledge acquisition, knowledge


representation, knowledge inference and knowledge transfer to user

Inference rules

Degrees or levels of expertise -> mimic human expert

Transferring expertise

Enable experts to make better and faster decisions than non-experts


in solving complex problem

Problem solving procedure -> knowledge stored at database

Explanation capability

Ability to explain its advice or recommendations ->


components/subsystem called justifier or explanation
subsystem

16

Fundamental Characteristic of an ES

Narrow specialized domain


High quality performance
Rules of thumb or heuristic
Explanation capability
Employ symbolic reasoning
Do not follow a prescribed
sequence of steps
17

QUALIFYING
CHARACTERISTICS
OF EXPERT SYSTEMS
ES are appropriate in domains when/where:

there are no established theories


human expertise is scarce or in high
demand, but recognized experts exist
the information is fuzzy, inexact or
incomplete
the domain is highly specific
18

QUALIFYING CHARACTERISTICS
OF EXPERT SYSTEMS Contd
ES are not suitable when/where:
applications are calculative or deterministic
in nature
a formula or model exists
human experts are plentiful
not integrable into existing systems
end users have to build the knowledge base

19

Designing ES
The process of building ES is called
Knowledge Engineering, consist of
three stages :
i.
Knowledge acquisition : the process of
getting the knowledge from experts
ii.
Knowledge representation : selecting
the most appropriate structures to
represent the knowledge
iii. Knowledge validation :testing that the
knowledge of ES is correct and complete
20

General Arch. Of ES

Working Memory

Knowledge Base

Inference Engine

I
N
T
E
R
F
A
C
E

21

ES ARCHITECTURE
Contd
Knowledge Base
Information - facts, rules, goals,
knowledge
Domain specific facts and heuristics from
experts are stored
Inference Engine
Workhouse of system
Implements inference mechanism
Carries out search through KB
Provides overall control of system

22

ES ARCHITECTURE
Contd
Working Memory
Keeps track of current problem
status and history of solution to date
An area in memory for storing problem
constructed from facts supplied by user or KB
User Interface
Consultation

interface - Query
facility
Knowledge acquisition facility
Graphical screen interface
Explanation

23

Types of ES

Ruled Based Expert System

Frame-Based System

Include several knowledge representation approach

Model-Based System

Representation of the object-oriented programming


approach

Hybrid System

Represented as a series of rules

Structured around the model that stimulates the


structure and function of the system under study

Ready-Made (Off-the-shelf) System

Custom-made, similar to application package such as


an accounting general ledger or project management
in operation mgmt.

24

How ESs Work - Structure of ES


Consultant Environment

Development Environment

25

Main Player in the ES dev. team


Project Manager

Domain Expert

Knowledge Eng.

Programmer

Expert system

End user

26

STAGES IN ES DEVELOPMENT

Identification and Definition of


Problem,
Conceptualization
Formalization
Implementation
Testing and Validation
Integration
Maintenance
27

STAGES IN ES DEVELOPMENT Contd

Identification
determination of the scope of the
problem and its characteristics
identify the experts, resources
needed, and goals of the ES
determine whether problem can be
addressed by an ES

28

STAGES IN ES DEVELOPMENT
Contd
Conceptualization

knowledge acquisition
deciding on possible decomposition of problem
into subtasks
study, in detail, nature of domain knowledge to
identify concepts required to arrive at a solution
of the proble
the control strategy for properly using these
concepts is considered
29

STAGES IN ES DEVELOPMENT Contd


Formalization
knowledge engineer selects suitable representation
scheme(s) for the domain knowledge

the knowledge is organized and formalized on paper


using knowledge nets or decision trees, so that
domain expert can also browse through to check for
inconsistency or redundancy

inference mechanisms are selected depending on


the flow of information and the underlying structure
of the domain
30

STAGES IN ES DEVELOPMENT
Contd
Implementation

an ES development tool of shell is


selected, which has the capabilities to
represent the formalisms and inference
mechanisms adopted in the previous
stage
the knowledge base is coded according to
syntax of shell interpreter
the code is entered into computer
31

STAGES IN ES DEVELOPMENT
Contd
Testing and Validation

the ES is made to run on a large number of


representative test cases and debugged
validation is done with help of domain expert

Integration

an often overlooked and unattended stage of


ES development
ES is integrated into existing systems or into
the day-to-day operations of the organization
32

Problem areas addressed by ES


Main areas of application of ES (waterman, 1996)

Interpretation drawing high-level conclusions based on


data
Prediction projecting probable outcome
Diagnosis determining the cause of malfunctions, disease,
etc
Design finding best configuration based on criteria
Planning proposing a series of actions to achieve a goal
Monitoring comparing observed behavior to the expected
behavior
Debugging and repair prescribing and implementing
remedies
Instruction assisting students in learning
Control governing the behavior of a system

33

Benefits of ES

Increased productivity find solutions much faster than


human
Availability of expertise human experts can be at one
place at one time
Can be used in dangerous environments e.g. in space
Decreased decision making time
Increased process and product quality
Reduced downtime
Ability to work with incomplete and uncertainty information
Improved decision making process
Provide training
Ability to solve complex problem

34

Problem & Limitation of ES

Difficulty in engineering, especially acquiring


the expertise
Mistrust by user
Effective only for narrow specific areas
area of expertise
Hard, even for highly skilled expert to
abstract good situational assessment when
he or she is under time pressure
Knowledge is not always readily available

35

Expert System Success Factors

A champion in management
User involvement
Training

36

SUCCESSFUL APPLICATIONS OF ES
IN ENGINEERING & RELATED FIELDS
Contd
Education

Expert general library reference (Drexel University)


Learning disability classification advisor
Speech pathology advisor
Student behavior consultant
Textbook selection advisor
Tutor designers in design checking (DECGUIDE,
Lockheed Sunnyvale)
Tutoring users of VMS operating system (Digital
Equipment Corporations

37

SUCCESSFUL APPLICATIONS OF ES
IN ENGINEERING & RELATED FIELDS
Contd

Ordering Systems and Marketing

Configures VAX orders (XCON, XSEL, AND


XSITE, Digital Equipment Corporation)
Order checking (OCEAN, NCR Corporation)
Order entry checking (Nixdorf)
Promotions of goods (PROMOTER, MDS)

38

SUCCESSFUL APPLICATIONS OF ES
IN ENGINEERING & RELATED FIELDS Contd
Sales

Guide to selling methods


Pricing of custom products
Sales support
Sales commission and bonus

Scheduling

Airline crew scheduling


Job shop scheduling (Westinghouse)
Optimal airline seating (Northwest Orient Airlines)
Production management system (ISIS)
Scheduling manufacturing orders (ISA

39

SUCCESSFUL APPLICATIONS OF ES
IN ENGINEERING & RELATED FIELDS Contd

Computers and Communications


Analyze telephone switching systems (COMPASS, GTE)
Check order entry (CONAD, Nixdorf)
Computer configuration (XCON, XSEL, and XSITE, Digital
Equipment Corporation)
Diagnose circuit fabrication lines (PIES, Fairchild)
Diagnosing failures in data processing equipment (DIAG8100,
Travellors)
Diagnosing failures in disk drive (FAULTFINDER, Nixdorf)
Diagnosing failures in tape drives (AI-SPEAR, Digital
Equipment Corporation)
Sequence steps in PC board assembly (HI CLASS, Hughes
Electro Optical & Data Systems)
Troubleshooting Ethernet networks (NTC, Digital Equipment
Corporation)
Troubleshooting communication hardware (BDS, Lockheed)

40

SUCCESSFUL APPLICATIONS OF ES
IN ENGINEERING & RELATED FIELDS Contd
Engineering
Carburetor fault diagnosis
Construction project planning and evaluation
Design of foundations for bridges and buildings
(Carnegie-Mellon University)
Fastener selection
Front end for engineering design package
Linear programming system (American
University)
Material handling equipment selector (North
Carolina State University)
41

SUCCESSFUL APPLICATIONS OF ES
IN ENGINEERING & RELATED FIELDS Contd
Manufacturing
Analysis and prevention of mechanical failures (Duke
University)
Chemical material selection
Continuous-process manufacturing advisor
Detecting cracks in billets (Kawasaki Steel)
Diagnosis of hydrostatic sterilizers (Campbell Soup)
Diagnosis of railroad locomotives (DELTA, General Electric)
Maintenance of epi-reactor (IMP, Texas Instruments)
Newspaper layout design (Composition Systems)
Optimum performance maintenance (Ingalls Shipbuilding)
Process control applications (PICON, Lisp Machine, Inc.)
Sequencing computer board assembly (HI CLASS, Hughes)
Troubleshooting circuits (Hewlett-Packard)

42

Expert systems
Define expert system.
What are the features of an expert system?
Write short notes on expert system building ?
How do expert systems differ from conventional programs ?
What are the advantages and disadvantages of artificial expertise over
human Expertise?
Discuss the basic characteristics of an expert system.
Describe a typical expert system architecture
What are the different types of knowledge representation. Explain them in
detail with examples.
Explain the difference between forward and backward chaining and under
what conditions each would be best to use for a given set of problems.
When is expert system development possible ? explain the tasks in
43
building expert systems.

Applications ( POWER SYSTEMS)


Assume that you are an expert engineer assigned with the task
of developing expert system package for a power system
problem. State the power system problem of your choice and
describe the step-by-step procedure of developing the ES
package.
Explain with a functional diagram, an intelligent alarm processor
for energy management system using a real time expert system.
Assume various system conditions like frequency deviation,
voltage deviation etc.. that generate alarms.
Expert system application for voltage and VAR control in largescale power system.
44

Вам также может понравиться