Organizing Data and Information

advertisement
Specialized Business
Information Systems
Chapter 11
Chapter 11 discusses specialized information
systems used in business, including artificial
intelligence, or AI, expert systems, and virtual
reality. After studying this chapter, you should be
able to address the objectives on the next 3 slides.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 2
Learning Objectives
Define “artificial intelligence” (AI)
State the objective of developing AI
systems
List the characteristics of intelligent
behavior & compare natural & artificial
performance for each
Identify the major types of AI systems &
provide an example of each
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 3
The field of artificial intelligence includes several
different types of systems that replicate or mimic
functions of the human brain. Although artificial
systems are better at some things than are
humans, humans surpass machines at others.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 4
Learning Objectives
List the characteristics and basic components
of expert systems
Identify at least 3 factors to consider in
evaluating the development of an expert
system
Outline & explain the steps in developing an
expert system
Identify benefits associated with expert
system use
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 5
Expert systems are a type of artificial intelligence
that is widely used in business. Expert systems
provide novices with the capabilities of an expert.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 6
Learning Objectives
Define the term “virtual reality” and give
three examples of virtual reality
applications
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 7
Virtual reality systems offer a new, highlyinteractive, three-dimensional interface between
computers and people. Virtual reality applications
have begun to spread through businesses.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 8
An Overview of Artificial
Intelligence
The term “artificial intelligence” was coined in the
nineteen fifties to describe computers with
capabilities that duplicated or mimicked the
functions of a human brain. At the time, some
predicted that computers would be as smart as
people within a matter of years. While
developments in AI have not met these optimistic
early expectations, they have been beneficial.
AI systems demonstrate characteristics of human
intelligence, replicating human decision-making
for certain problems.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 10
Since AI focuses on replicating intelligent human
behavior, it helps to understand the nature of
intelligence. Unfortunately, intelligence is not easily
defined. The next 2 slides contain at least a partial
description.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 11
The Nature of Intelligence
Learn from experience & apply the
knowledge
Handle complex situations
Solve problems when important
information is missing
Determine what is important
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 12
Humans naturally learn from experience – that is,
by trial and error. And humans can apply what they
learn to different contexts. Neither trait comes
naturally to AI systems – they can only learn and
apply what they’ve been programmed to learn –
and the programming is difficult. Humans can learn
in multiple areas and automatically apply what they
learn.
Humans are often involved in complex situations.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 13
In business, executives face complicated
legislation, rapidly changing markets and
competition, and many more complexities, yet must
make decisions, sometimes quickly, that affect their
company’s future. People can make mistakes and
learn from them. However, computers can handle
only those complex situations that they’re
programmed to handle.
Humans continually make decisions under
uncertainty – that is, with partial or even inaccurate,
information. AI systems can handle such situations
in many contexts.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 14
Everyday humans receive masses of incoming
information. People can screen the information
and discard irrelevant information – a skill built
through experience. Computers are limited by their
programming – and it’s not easy to program a
computer to know what’s irrelevant.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 15
The Nature of Intelligence
React quickly & correctly to new
situations
Understand visual images
Process & manipulate symbols
Be creative & imaginative
Click here for a
Use heuristics
computer’s poem. Is it
creative?
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 16
Humans experience “gut instincts” – like when you
walk down a street and “know” you should leave or
get hurt. Children know not to touch a flame.
Computers have no gut feelings, and can only
react quickly to specific stimuli for which they’re
programmed.
Even state of the art computers have trouble
interpreting visual images. When a person sees
their reflection in the mirror, they know it’s a
reflection and not a clone. When you see people,
you look for many clues in dress, grooming, and
behavior to determine their gender.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 17
Walking down a crowded street is natural even for
children. But none of this is natural to computers.
Research in the area of perceptive systems – that
is, machines that can mimic human hearing, sight,
or touch – has progressed and, as we’ll see later in
the chapter, some systems have limited recognition
ability.
Although computers excel at rapidly processing
numbers, they’re not so good at processing visual
information. Again, they’re limited by their
programming.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 18
Humans can think of new products and services
and create novel objects. Although computers
have been used to write poetry or draw, few can
yet be considered truly creative.
Heuristics are rules of thumb developed through
experience. People often use heuristics in
decision-making. For instance, if you leave home
for work after 7:30 AM, you may choose an
alternate route, since experience has shown that
by then there is often an accident backing up traffic
on your normal route.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 19
Or you may ignore the weather forecast of
precipitation if the chance is less than 70%,
because based on your experience, it rarely rains
or snows unless the chance is 70% or higher.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 20
Figure 11.1
Deep Blue is an example of a computer
programmed to learn from past chess moves.
However, that’s all it can learn. Currently
computers can only learn what and how they’ve
been programmed.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 22
Table 11.1
Table 1.1 summarizes the strengths and
weaknesses of natural and artificial intelligence.
Computers can do many things that humans
consider difficult – such as making complex
calculations – but are poor at tasks people consider
easy – such as processing images and learning
from experience. However, as research on human
intelligence and AI research continues, the gap
between computer & human capabilities is
shrinking. Some researchers believe the gap will
be closed in the 21st century.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 24
Figure 11.2
The field of AI includes expert systems, robotics,
vision systems, natural language processing,
learning systems, and neural networks. Some of
these areas are interrelated. Expert systems are
hardware and software that can solve problems
based on the knowledge of a human expert.
Because of their importance in business expert
systems will be covered later in detail. The other
areas of AI will be briefly addressed in the next few
slides.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 26
The Major Branches of
Artificial Intelligence
Vision systems
Learning systems
Neural networks
Robotics
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 27
Vision systems include hardware and software that
enables computers to capture, manipulate and
store visual images and pictures. Examples of uses
of visual systems include fingerprint identification,
identifying people based on facial features, and
identifying geographic features from satellite
images. Vision systems can also be used for quality
control purposes in manufacturing. Robots with
vision systems can negotiate there way around
obstacles.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 28
Natural language processing allows computers to
understand “normal” language, such as English or
German – that is, language like it is normally used,
not commands created according to a specific
programming syntax. Natural language processing
is the most complex form of voice recognition. Voice
recognition allows a system to recognize commands
or take dictation. There are products available now
to do this, after they have been trained. Most require
brief breaks between words to be accurate or they
are limited to a specific topic.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 29
For instance, voice recognition is used over the
phone instead of menus requiring you to press a
number on the phone number pad. Voice
recognition has been used with some success to
make airline reservations. Continuous voice
recognition is when a system recognizes natural
speech – and understands the meaning. This
natural language processing is difficult because
the system must not only understand idioms or
accents, but must also understand context.
Consider, for example, the passage “Jane saw a
letter on the desk. She quickly read it.”
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 30
Humans think nothing about what the sentence
means – its clear to us that Jane quickly read a
letter she found and the letter was on a desk. But
to artificial intelligence, there are ambiguities that
can only be resolved by understanding context and
norms. For example, “it” could refer to either the
letter or the desk. Or Jane may have seen a letter
while she was on a desk. People know humans
usually read letters, not desks, and that generally
letters, and not people, are on desks. But unless
they’ve been explicitly programmed to, computers
don’t possess such “common sense” knowledge.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 31
Researchers are still a long way from developing
artificial intelligence that can understand natural
language.
Learning systems include hardware and software
that allows a system to “learn” from the results of its
actions or environment – that is, learn from
experience. Learning systems require feedback on
the results of their actions or decisions and whether
those results are good or bad. Deep Blue, IBM’s
chess-playing computer, is a learning system. It
learns from its actions and their consequences, as
well as the actions of its opponent.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 32
However, unlike humans who can learn from
experience across a broad range of areas, this is all
Deep Blue can learn. It couldn’t learn, for example,
that when its opponent scratches it head, he feels
vulnerable and makes a careless move.
Whereas expert systems try to model an expert’s
thought processes in software, neural networks try
to model the brain itself in hardware. Neural
networks use massively parallel processors in an
arrangement based on the composition of the
brain’s neurons.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 33
Strengths of connections between neurons
continually adjust in a human brain as a result of
learning. The electrical strengths between
processors in a neural network readjust to codify
learning. For example, a neural network could be
given profiles of “normal” credit card charge activity
and fraudulent activity. After being given the details
and told if this was normal or fraudulent, the neural
net would readjust strength between processors.
After being trained on a large number of examples,
the neural network could identify fraudulent credit
card use on new cases.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 34
Neural nets model the human brain’s ability to
process many pieces of data at once and learn to
recognize patterns in the data. Pattern recognition
is useful for trend analysis, data mining, solving
complex problems based on partial information, and
quickly updating data. Although the most powerful
neural nets are hardware-based, software is
available to create neural nets on standard
computers.
Robotics involves using computer or mechanical
devices that can perform "physical" human tasks
that require precision or are dangerous for humans.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 35
Figure 11.3
Many robots today are specialized for a single
function, such as mounting tires on an assembly
line, and are not the android-like robots of science
fiction fame. The software controlling the robot is
critical today, since processing power of the
average robot gives it the brain capability of an
insect. To approach human intelligence, the robot
brain would need to perform about 100 trillion
operations per second – it now does about 10
million. However, some researchers believe that
processors will achieve that speed in the first part of
the 21st century.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 37
Biomedicine/biotechnology sometimes involving
implanting robotic devices in the human body, such
as artificial vision or robotic limbs.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 38
An Overview of Expert
Systems
Expert systems contain the knowledge of an expert
in a specific area and use that knowledge to
replicate human problem solving in that area. Like
human experts, expert systems draw inferences
from the rules, facts, and relationships in their
knowledge base and use heuristics to draw
conclusions or make recommendations. Expert
systems exist to diagnose problems, predict events,
plan, and design new products and systems. For
example, expert systems can be used by help desk
personnel to troubleshoot problems end users have
with software.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 40
Expert systems can be used to configure a complex
computer system. Expert systems have been used
in business to reduce costs, increase profitability,
explore business options, and improve customer
service.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 41
Figure 11.4
One use of expert systems in business is to
determine credit limits for credit cards.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 43
Characteristics of an Expert
System
Can explain reasoning
Can provide portable knowledge
Can display “intelligent” behavior
Can draw conclusions from complex
relationships
Can deal with uncertainty
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 44
An expert system can explain how it reached a
conclusion by showing the path of rules and
inferences in its knowledge base that it followed.
This is valuable to users of the conclusions. For
instance, a physician using an expert system to
help diagnose a blood disease could compare the
expert system’s reasoning to her own to determine
her level of confidence in the system’s conclusion.
This is also useful in training novices in an area.
For example, a new loan processor making a
decision to approve or deny a loan can see the
expert system’s reasoning and learn from it.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 45
Because an expert’s knowledge is codified in an
expert system, expert systems can preserve scarce
expertise and give others access to it.
Given a data set, an expert system can propose
new ideas, which is a characteristic of expert
behavior. For example, expert systems can
diagnose patients’ conditions from their symptoms
or suggest where to drill for oil, based on geologic
data and expert knowledge.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 46
Expert systems can evaluate complex
relationships to reach a conclusion or make a
recommendation. Although expert systems
generally require a well-structured problem, it can
have many complex relationships. The information
can be incomplete or somewhat inaccurate, since
expert systems can use probabilities and
heuristics.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 47
Limitations of Expert Systems
Limited to narrow problems
Not widely used or tested
Hard to use
Cannot easily deal with “mixed”
knowledge
Possibility of error
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 48
Generally, the narrower the scope of a problem, the
easier it is to build an information system to solve
it. For example, a medical expert system can
diagnose a particular category of diseases, say,
skin diseases, but cannot advise a person on the
amount of exercise he should get a week or
whether the medication for the skin disorder will
interact other medications the patient takes.
Because of this narrow focus, an expert system is
not typically used in numerous organizations and
thus, is not widely tested.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 49
Some expert systems require a user to have a
technical person help them use it. As with all kinds
of information systems, user friendliness is
important and should be a priority for designers of
expert systems.
There are several ways that knowledge can be
represented in an expert system’s knowledge base.
For example, knowledge can be defined by rules or
by comparison to case scenarios. Normally, only a
single method can be used within a particular
system’s knowledge base.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 50
Since the main source of an expert system’s
knowledge is a human expert, the knowledge could
be incomplete or incorrectly documented in the
knowledge base. Since developing expert systems
is very complex, there is also an opportunity for
programming error.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 51
Limitations of Expert Systems
Cannot refine own knowledge base
Hard to maintain
Possible high development costs
Raise legal & ethical concerns
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 52
If an expert system is to learn from experience, it
can only learn what it is been programmed to learn.
Deep Blue is an example of such an expert system.
Many expert systems cannot refine or maintain their
own knowledge bases – for instance, they cannot
eliminate redundant or contradictory rules. Since
they are not generally self-adapting, maintaining an
expert system is difficult and labor intensive.
Because of the complex relationships represented
in the knowledge base, it is sometimes too difficult
or costly to change the rules and relationships.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 53
Because expert systems involve programming
complex relationships, developing an expert system
can be labor intensive and costly. Expert system
shells can be used to streamline development and
reduce costs. An expert system shell consists of
tools that can be used to develop an expert system
and much of the logic required to search through a
knowledge base.
Legal and ethical questions continue to surround the
use of expert systems.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 54
For example, when a patient dies because of a
physician’s negligence, legal remedies can be
taken. However, if a patient dies because of an
action suggested or taken by an expert system,
who is responsible? The physician? The software
developer? The experts who provided the
knowledge for the expert system?
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 55
Fig 11.5
Expert systems have been successfully used in all
stages of problem-solving and in a broad range of
disciplines. Figure 11.5 shows some of the primary
uses of expert systems.
An expert system can help managers evaluate and
select priorities or goals, by analyzing the firm’s
strengths, weaknesses, and opportunities, as well as
those of its competitors.
Some of the earliest uses of expert systems were for
diagnosis and design.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 57
For example, an expert system can configure a
complex computer installation more quickly &
accurately than can a human. Expert systems
have been used to diagnose particular types of
diseases, such as skin disorders.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 58
Fig 11.6
Expert systems can diagnose problems, or
potential problems, of machinery or operations.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 60
When to Use Expert Systems
High payoff
Preserve scarce expertise
Distribute expertise
Provide more consistency than humans
Faster solutions than humans
Training expertise
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 61
Since expert systems can be difficult and expensive
to develop, they should be used where they can be
most beneficial. This slide summarizes situations
where expert systems have been shown to be
worth implementing.
Clearly, when there is a high potential payoff, or
when the expertise is needed at a place dangerous
to humans, it makes sense to develop the expert
system.
It is generally also worthwhile to develop an expert
system to capture and preserve expertise that not
many people have, that is expensive, or that can’t
be duplicated in other ways.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 62
Also, an expert system is called for when this kind
of scarce expertise is needed in many locations at
once.
No matter how hard they try, people cannot be
100% consistent – they tire, have bad moods, or
are distracted. Where consistency is needed – say
in loan approval – investing in an expert system
may be worthwhile.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 63
In complex tasks, such as configuring large
computer installations, it may take humans too long
to do the job for the company to be competitive.
Using an expert system to complete the task
quicker than your competition would be wise.
And finally, sharing scarce expertise or training
others in the area, is a solid use of expert systems.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 64
Fig 11.7
An expert system is a collection of integrated
components, including a knowledge base, an
inference engine, an explanation facility, a
knowledge base acquisition facility, and a user
interface. As shown in Figure 11.7, the user
interacts with the user interface, which interacts
with the inference engine. The inference engine is
central and interacts with all components.
The knowledge base stores all relevant data,
information, rules, cases, or relationships that the
expert system uses. Each application has a unique
knowledge base, analogous to an expert’s store of
information and experience.
Principles of Information Systems,
Chapter 11
Fifth Edition
Slide 66
Consider an expert system to diagnose mechanical
problems with a car. The knowledge base would
include an expert mechanic’s knowledge about the
major car systems, symptoms of failure of the
various symptoms, components of the systems,
parts that normally fail at the same time, and so on.
The knowledge base would also include a
mechanic’s knowledge about the steps to use in
diagnosing a problem and the relationships
between different kinds of symptoms. Expert
systems can also be integrated with other
information systems by sharing a common
database.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 67
Eliciting such information for the knowledge base
from experts is a challenging task. First, it is not
easy for someone to be explicit about their
expertise and actions. For example, when asked
how she diagnoses spark plug problems, a
mechanic’s first response might be “because the
car sounds like it has spark plug problems”. Also,
since expert systems typically represent a
compilation of multiple experts’ input, often experts
don’t agree on relationships or data. It’s up to the
designers of an expert system to elicit detailed
information from the experts and determine what to
include in the knowledge base.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 68
The Knowledge Base
Rules
Cases
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 69
Information in a knowledge base can be structured
in different ways. Rules and cases are two ways to
organize a knowledge base.
A rule is a conditional statement that relates
conditions to specific outcomes, using if-then-else
constructs. Rules are often combined with
probabilities. For example, a rule in a weather
forecasting expert system might be “If the dew point
is greater than 90%, then there is a 60% chance of
rain.”
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 70
When an expert system uses cases, it addresses
its problem by finding a case in its knowledgebase
similar to the situation at hand, and modifying the
cases outcomes to accommodate the current
situation.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 71
Fig 11.8
Figure 11.8 shows a rule that could be in the
knowledgebase of an expert system to approve or
disapprove loan applications. Notice the “if-thenelse” rule structure.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 73
Inference Engines
Backward chaining
Forward chaining
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 74
The inference engine searches the knowledgebase
for information and relationships relevant to the
current problem, and provides the predictions,
diagnosis, or suggestions. The inference engine
must find the right pieces of the knowledge base
and assemble them correctly. Two of the ways
inference engines work is by backward or forward
chaining.
An inference engine uses backward chaining if it
starts with conclusions and works backwards to the
initial facts. If the facts don’t support the
conclusion, it tries another conclusion and works
back to its initial facts.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 75
That is, it starts with the “then” side of the rule.
An inference engine uses forward chaining if it
starts with facts and works forward to reach a
conclusion, starting with an “if” clause. Forward
chaining is often used by costlier expert systems.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 76
Fig 11.8
For example, in Figure 11, an inference engine
using backward chaining would start with the
premise “Accept loan application.” It would then
look through the facts it knows about the applicant,
or ask the loan officer for more information. For
example, it would check if the applicant had prior
credit problems, net income at least 4 times greater
than the monthly loan payment, and so on. If the
applicant met all the conditions, the system would
recommend approval. If the applicant failed to meet
one of the conditions, the expert system would
repeat the process using a different rule.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 78
A forward chaining system, on the other hand,
would start with facts about the applicant. If credit
history was okay, then the systems would check net
income, and so on. If the applicant met all the
conditions, approval would be recommended. If he
failed a condition, the expert system would go on to
a different rule and start checking conditions again.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 79
The explanation facility allows an expert system to
explain the line of reasoning finally used to the
system’s user. For example, if the loan officer
asked for an explanation of an approved loan, the
expert system would say “There are no previous
credit problems and monthly income is 4 times
monthly loan payment” and so on.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 80
Fig 11.9
The knowledge acquisition facility provides an easy
way to input data and rules into the knowledge
base. Expert system shells generally have a simple
user interface of menus and forms for input. The
information is then stored in the correct format in
the knowledge base. Knowledge acquisition can
also be partially or completely manual. In a custom
expert system written in a programming language
without using an expert system shell, knowledge
acquisition can be tedious.
The user interface component allows users of IS
professionals to create, modify and use an expert
system.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 82
Fig 11.10
Like other information systems, expert systems are
best developed using a systematic approach, as
shown in Figure 11.10. First, the objectives and
requirements for the expert system are determined.
Next, experts are identified. Finding experts to
participate is not necessarily easy. Since expert’s
possess rare skills or knowledge, they’re generally
needed to do the work they’re expert in. And
participating in expert system development can
require a lot of time away from their normal work.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 84
Expert systems often must be constructed by
highly skilled personnel, so consultants may need
to be hired. After the expert system is put into
use, it must be evaluated, monitored, and
maintained.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 85
Fig 11.11
Often more than one type of participant is required
to develop an expert system. The specific area of
knowledge that the expert system will address is
called a domain. A domain expert is the people or
group who has the skills or knowledge to be
captured in the expert system. The knowledge
engineer(s) is the IS professional who has been
trained to design, develop, and implement expert
systems. Knowledge users are the people who
will ultimately use the expert system.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 87
Fig 11.12
Although in theory, expert systems can be
developed using any programming language,
specialized tools have been developed to make it
easier. In the 1980s, LISP and Prolog, two
specialized AI languages, were often used to
develop expert systems. More recently, expert
systems shells makes development even easier.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 89
Table 11.2
Expert system shells exist for all types of computer
platforms – from PCs to mainframes. Table 11.2
describes some popular ones. After an expert
system has been developed, it can be used by
people with little or no computer experience. The
expert system will ask the user questions to get the
information it needs to reach a conclusion.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 91
Advantages of Expert
Systems Shells and Products
Easy to develop & modify
Use of satisficing
Use of heuristics
Development by knowledge engineers
& users
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 92
Expert systems developed with traditional
programming languages or with AI languages such
as LISP and PROLOG are difficult and costly to
maintain and modify. Shells have an editing facility
that simplifies this process.
The traditional approach to problem solving
involves finding the best solution. Advanced
languages and tools return good, though not always
optimal, solutions. This decreases development
time and expense, as well as decreasing the time
required for the system to find a solution.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 93
Heuristics, or rules of thumb, are useful in
satisficing. Heuristics are often easier to write using
an expert system shell than they are using a
programming language.
In addition to knowledge engineers, systems
analysts and programmers are needed to build an
expert system using programming languages, which
can be costly. If an expert system is used,
knowledge engineers can serve more as
consultants to users in building the expert system.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 94
Figure 11.13
Figure 11.13 shows alternative ways to develop
expert systems and their relative costs and time.
In-house development from scratch is the costliest
and most difficult alternative. However, if an
organization requires a highly customized system
and needs more control over its features, it’s a
sound development approach. Using an expert
system shell is usually faster and less costly, but
the resulting system may not be exactly what is
wanted.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 96
Buying an existing expert system package is the
easiest and fastest way to acquire an expert
system, especially when an organization needs a
relatively standard system. However, expert
systems packages may not satisfy unique
requirements.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 97
Applications of Expert
Systems & AI
Credit granting
Shipping
Information management & retrieval
Embedded systems
Help desks & assistance
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 98
Throughout the chapter we have mentioned uses of
expert systems. Several are listed on this slide.
Expert systems are starting to be embedded in
larger systems where we don’t expect them. For
example, the antilock braking system on cars is an
expert system.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 99
Some physicians use expert systems to determine
a patient’s likelihood of having particular diseases.
MatheMEDics’ Easy Diagnosis is an example of an
online medical expert system.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 101
Virtual Reality
A virtual reality system allows one or more users to
interact with the system in a computer-simulated
environment.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 103
Fig 11.14
Special hardware interface devices are needed to
interact with the simulated world through sight,
sound, and sensation. Users generally wear a
head-mounted display to view the virtual world.
The head mounted display tracks the location of
the user’s head & where the user is looking to
continually change the display the user sees as
she move her head. Users can hear sounds
through earphones. The least developed interface
is the haptic– that is, the sense of touch. Although
users can wear special gloves to manipulate
objects in a virtual world, researchers are still
working on relaying sensations back to the user.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 105
Fig 11.15
In immersive virtual reality, the users can interact in
a full scale three-dimensional environment. Figure
11.15 shows an immersive virtual reality system,
where users are exploring the Detroit Midfield
Terminal.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 107
As the cost of technology decreases, virtual reality
applications will become more widely available.
Virtual reality applications can be found in medicine,
real estate, education, and entertainment. Figure
11.16 shows a computer-generated image used in
sports simulations and special effects in movies.
Chapter 11
Principles of Information Systems,
Fifth Edition
Slide 109
Download