You are on page 1of 15

Visit: www.geocities.com/chinna_chetan05/forfriends.

html

EXPERT SYSTEMS
Abstract:
Expert systems have proven to be effective in a number of

problem domains, which normally require the kind of intelligence

possessed by a human expert the kind of intelligence possessed by a

human expert. The areas of application are almost endless. Where

ever human expertise is needed to solve problems, expert systems are

likely candidates for application.

Application domains include aerospace, banking, biology,

chemistry, engineering, finance, geology, geophysics, law,

manufacturing, meteorology, military operations and more.

In this paper I explored Expert system Architecture with Key

roles, building tools and shells, continued with how to represent

knowledge for expert systems followed by its features and explained

briefly about Expert system with an example and the next generation

expert systems and at last discussed about pitfalls and limitations of

expert system.

1 Email: chinna_chetan05@yahoo.com
Visit: www.geocities.com/chinna_chetan05/forfriends.html

Index

1. Introduction

2. Expert system Architecture

 Key roles in Expert system.

 Steps in building

 Expert system shells

3. Representing knowledge and languages for expert system

4. Features of expert system

5. Brief Description of Advanced Expert system examples

6. Upcoming Expert Systems in Next Generation

7. Pitfalls & Limitations of Expert systems

8. Conclusion

9. Bibliography

2 Email: chinna_chetan05@yahoo.com
Visit: www.geocities.com/chinna_chetan05/forfriends.html

INTRODUCTION
Right from the evolution of computers, man is trying to lessen
his burden and started in calculating all his abilities and natural
possession. He was gifted with into this Magic machine -The Computer.
In 1970’s, the beautiful thought of making the machine intelligent
emerged out of the supreme computer- The Human Brian, the result
that is our present topic - Expert Systems.
ARTIFICIAL INTELLEGENCE:
Artificial Intelligence broadly defined as concerned with the
intelligent behavior in artifacts.
Artificial Intelligence have always been surrounded by
controversy Question of Interest raised here is,
"CAN MACHINES THINK?"
Somewhat supporting and hope raising answer is that, "A virus
can be called as a Machine One made of proteins".
Finally it's been modified to, "The study of how to make
computers do things which at the Moment, people do better".
EXPERT SYSTEMS:
An expert system is a computer program using expert
knowledge to attain high levels of performance in a narrow problem
area. They are the capable programs that contained the knowledge
requirements to mimic expert human performance at several tasks,
including diagnosis, design and analysis.

EXPERT SYSTEM ARCHITECTURE:

3 Email: chinna_chetan05@yahoo.com
Visit: www.geocities.com/chinna_chetan05/forfriends.html

PLAYERS IN EXPERT SYSTEM GAME

The main players in the expert system are the:

Expert System: An expert system is a computer program using
expert knowledge to attain high levels
performance in a narrow problem area.

Domain Expert: He is an articulate, knowledge person with a
reputation for producing good solutions to
problems in a particular field.

Knowledge Engineer: He is a human, usually with a background in
computer Science and AI, who knows how to
build expert systems.

Expert System Building Tool: It is the programming language used by
the knowledge engineer or programmer to
build the Expert system.

The user: He is the human who uses the expert system
once it is developed.

STEPS IN BUILDING AN EXPERT SYSTEM
IDENTIFICATION: How can important aspects in task be
recognized?
CONCEPTUALIZATION: What concept is needed to produce a
solution?
FORMALISATION: How can the knowledge be formally
represented?
IMPLEMENTATION: What rules will embody the knowledge?
TESTING: How can the rules be validated?

EXPERT SYSTEM SHELLS:
In general the systems were constructed as a set of declarative
representations (rules) Combined with an interpreter for those rules, it
was possible to separate the interpreter from the domain-specific
knowledge and thus to create a system that could be used to construct
new expert systems by adding new knowledge to the new problem
domain. The resulting interpreters are called as "Shells".

Ex: - EMYCIN is the SHELL for the famous expert system; this
shell was made by removing the medical knowledge base of THE
EMYCIN expert system.

4 Email: chinna_chetan05@yahoo.com
Visit: www.geocities.com/chinna_chetan05/forfriends.html

REPRESENTING KNOWLEDGE AND LANGUAGES FOR EXPERT
SYSTEM:

FRAME WORK - The Organization:

Knowledge, mean the information a computer program needs
before it can behave Intelligently. This knowledge should undergo two
phases. They are:

Organizing Knowledge: -

Many of the rules in expert systems are heuristics- rules of
thumb or simplifications that effectively limit the search for solutions.
Expert systems use heuristics because the tasks these systems
undertake, such as finding new mineral deposits or settling a lawsuit,
are typically difficult and poorly understood. An algorithmic method
guarantees to produce the correct or optimal solution to a problem,
while a heuristic method produces an acceptable solution to most of
them.

The knowledge Base in an expert system contain,
 Facts
 Rules
The "Inference Engine" contains an interpreter that decides
how to apply the rules to infer new knowledge.

REPRESENTING KNOWLEDGE:

The three most widely used representations used in current
expert systems are,

 Rules
 Frames
 Semantic Nets

1. RULE: A formal way of specifying a recommendation, directive,
expressed as,
"IF premise THEN conclusion".
Rule based knowledge representation centers on the use of
IF condition THEN action statements.
Ex: - if the patient was an insulator before 1965, then the
patient
directly handled asbestos.
This sort of matching is what is called "INFERENCE CHAINS".

5 Email: chinna_chetan05@yahoo.com
Visit: www.geocities.com/chinna_chetan05/forfriends.html

2. FRAME: A knowledge representation method associated
features with nodes representing concepts or objects.
Frame based knowledge representation uses a network of nodes
connected by relations and organized into a hierarchy. Each node
represents a concept that may be described attributes and values
associated with the node. Nodes low in the hierarchy automatically
inherit the properties of higher-level nodes.
3. SEMANTICNET: The part of the representation method consisting
of a network of nodes standing for concepts or objects,
connected by arcs describing the relations between the nodes.

LANGUAGES FOR EXPERT SYSTEM APPLICATIONS:

The programming languages used for expert system applications
and generally either Problem-oriented languages, such as FORTRAN
and PASCAL, or symbolic-manipulation languages, such as LISP and
PROLOG.

KNOWLEDGE ENGINEERING LANGUAGES: A knowledge
engineering language is a sophisticated tool for developing expert
systems. Knowledge engineering languages can be categorized as
either Skeletal Systems or General-purpose systems.

Ex: - The prospector has been stripped into KAS, a skeletal
system.

SUPPORT FACILITIES: The support facilities consist of tools for
helping with programming, such as debugging aids and knowledge
base editors, and tools that enhance the capabilities of the finished
system, such as built-in input/output and explanation mechanisms.

DEBUGGING AIDS: Most programming languages contain tracing
facilities and break packages.

IO FACILITIES: Different tools deal with input/output in different ways.

Some provide Run time knowledge acquisition.
Ex: - EMYCIN programs ask the user for needed information
whenever they can’t find it in the knowledge base.

FEATURES OF EXPERT SYSTEMS:

1. CORPUS OF KNOWLEDGE: The heart of expert system is the
powerful corpus of knowledge. The accumulation and codification
of knowledge is one of the most important aspects of an ES.

6 Email: chinna_chetan05@yahoo.com
Visit: www.geocities.com/chinna_chetan05/forfriends.html

2. HIGH LEVEL EXPERTISE: This feature is provided to aid in
problem solving. The expertise can represent the best thinking of
the top expert in the field, leading to problem solutions that are
imaginative, accurate and efficient.
3. PREDICTIVE MODELLING: The system can act as an
information processing theory or model of problem solving in the
given domain, providing the desired answers for a given problem
situation and showing how they would change for new situations.
4. TRAINING FACILITY: This feature is provided for key personnel
and important staff members. Expert systems can be designed
to provide such training, since they already contain the
necessary knowledge and the ability to explain their reasoning
processes.
DETAILED EXPLANATION OF AN EXPERT SYSTEM-
“PROSPECTOR”: -

Application Area: - GEOLOGY

Categories Covered under ES Applications: -
INTERPRETATION
DIAGNOSIS
DEBUGGING

History:- Expert system work in Geology started with Prospector,
a system developed by Stanford Research Center in 1970.
Prospector was designed to help Geologists locate ore deposits and
accurately predict the existence of a Multi million dollar
molybdenum deposit in 1980.

Effort Taken: -

 Prospector was implemented directly in INTERLISP, a powerful
but relatively low-level language as far as expert-system building
tools go.
 A sophisticated support package was developed for PROSPECTOR
that included both explanation and knowledge acquisition
facilities.
 The system needed extensive domain knowledge.

PROSPECTOR AT WORK:

Let’s discuss this with an example.
Ex:
 An exploration geologist starts by telling the program the
Characteristics of a particular prospect of interest-the geological

7 Email: chinna_chetan05@yahoo.com
Visit: www.geocities.com/chinna_chetan05/forfriends.html

setting, structural controls, and kinds of rocks, minerals and
alteration products present or suspected.
 The program compares these observations with models of
various kinds of ore deposits, noting the similarities, differences,
and missing information.
 The program then engages the geologist in a dialog to obtain
additional relevant information and uses that information to
assess the mineral potential of the prospect.
 Our goal is to provide the geologist with a service comparable to
giving him telephone access to authorities on many kinds of ore
deposits.

Rules Involved in the Above Example: -

IF: magnetite or pyrite in disseminated or vein let form is present
THEN:(2, -4) there is favorable mineralization and texture for the
propylitic stage.
Indications:‘2’  ‘2’ indicate that the presence of evidence is mildly
encouraging.
‘-4’ ‘-4’ indicates that the absence of evidence is strongly
discouraging.

BRIEF DESCRIPTION OF SOME ADVANCED EXPERT SYSTEMS:

CONDITION MONITORING: The basic idea being to provide the
process operators and shift engineers with a tool that on a 24-hour by
365 days basis performs health-checks on the plant in order to:
 Identify any inability of the plant to perform at the optimum
efficiency level
 Present and inform about the extent of the inability
 Describe the reason for such inability
Thus, the Condition Monitoring System performs a task
impossible for any human being. CMS is currently being
developed for Nuclear Power Plants and is under configuration
for many other plants.

GEOMET- A LEADER IN FORMULATING SOLUTIONS TO GLOBAL
INTIATIVES.

GEOMET ’s meteorology staff, in conjunction with artificial
intelligence group, has designed and developed several pc-based
expert systems to aid weather forecasters and planning personnel.
GEOMET was an early leader in AI, developing one of the first expert

8 Email: chinna_chetan05@yahoo.com
Visit: www.geocities.com/chinna_chetan05/forfriends.html

system decision aids for forecasting. GEOMET has researched AI
applications in all areas of environmental sciences, water management
and flood control, hazardous waste investigations, and neural
networking applications.

CHEMICAL DEMILITARIZATION:
Special Project’s division supports the monitoring operations,
and safety issues associated with chemical demilitarization, and the
operation of demilitarization equipment and systems required to
handle, store, destroy transport, or dispose of recovered non-stockpile
CWM and other toxic, hazardous or lethal substances. Their services
include:
• Meteorological and air pollution modeling and monitoring
• Personnel protective equipment selection, design, and
development
• Environmental health and safety support
• RCRA/ hazardous waste analysis

PERSONAL PROTECTIVE EQUIPMENT:

Personal protective equipment experts provide a full range of
services to help define the specific level of protection required in any
hazardous situation, then support client requirements form initial
concept through final product development and delivery.

PPE ensures performance through stringent testing, heat stress
evaluation human comfort engineering, reliability qualification,
environmental limits, and respiratory protection.
EXPERT SYSTEM WORK AT UNIVERSITIES:

• The main fields of interest for expert systems in universities are
human problem solving developing and applying information
processing techniques to model human cognition and memory.
• Each rule would say, “If I recognize some situation S in short-
memory,
Then I take some action A”.
• The process of rule – evocation and memory modification to
model human problem solving and called the resulting system a
“Production System”;

EXPERT SYSTEMS WORK AT RESEARCH ORGANISATIONS:
• The research organizations engaged in expert system work
come in a variety of types ranging from experienced leaders in
AI to bold newcomers
9 Email: chinna_chetan05@yahoo.com
Visit: www.geocities.com/chinna_chetan05/forfriends.html

• Some are large corporations with an AI division or group. Others
are small companies devoted almost exclusively to AI and
expert system applications.

EXAMPLES OF HIGH PERFORMANCE EXPERT SYSTEMS:

In Research Field: -
• The SYNCHEM2 expert system synthesizes complex organic
molecules without assistance or guidance from a chemist. The
system uses knowledge about chemical reactions to generate a
plan for creating the target molecule from basic building block
molecules.
• The DENDRAL expert system infers the molecule structure of
unknown compounds from mass spectral and nuclear magnetic
response data.
• The MACSYMA expert system performs symbolic manipulation of
algebraic expressions assisting the user with problems involving
limit calculations, symbolic integration, solution of equations,
canonical simplifications, and pattern matching.

IN BUSINESS:
• YES/MVS, developed by IBM, helps the computer operator
monitor and control the MVC
• DELTA was developed to help maintaining personnel diagnosis
and repair malfunctions in diesel electric locomotives.
• ACE provides troubleshooting and diagnostic reports for
telephone cable analysis.

EXPLANATION FACILITIES: Almost all expert systems can explain to
users how they reach particular conclusions, but not all provide the
same degree of software support explanation.
Ex: MYCIN, have a complete explanation mechanism built into
the tool itself, so that any expert system written in that language can
automatically access the mechanism.
The most common type of explanation mechanism deals with
Retrospective reasoning; it explains how the system reached a
particular state. They also handle Hypothetical Reasoning, where the
system explain what would have happened differently if a particular
fact or rule had been different, and counter factual reasoning why an
expected conclusion was not reached.

STAGES IN THE DEVLOPMENT OF EXPERT SYSTEM TOOLS:

EXPERIMENTAL SYSTEMS: Created for a specific task; the
developer seldom support this.

10 Email: chinna_chetan05@yahoo.com
Visit: www.geocities.com/chinna_chetan05/forfriends.html

RESEARCH SYSTEMS: They may be extensively tested and
may be supported by developer. But slow and inefficient.

COMMERCIAL SYSTEMS: They polished, streamlined, well
supportive and fast.

REAL TIME EXAMPLE: EMYCIN

TOOL: EMYCIN
TYPE: Skeletal system
FEATURES: Rule-based, Backward chaining, Certainty handling,
Explanation.
IMPLEMENTATION LANGUAGE: INTERLISP
DEVELOPER: Stanford University
RULE: IF the material composing the substructure is one of
metals and the analysis error that is tolerable is between 5 and
30, and The non dimensional stress of the substructure is
greater than 90
THEN it is definite that fatigue is one of the stress behavior
in
Substructure
ACTUAL EMYCIN RULE:
PREMISE: ($AND (SAME CNTXT MATERIAL (LIST OF METALS))
(BETWEEN * CNTXT ERROR 5 30)
(GREATERP * CNTXT NO_STRESS 0.9)
(BETWEEN * CNTXT CYCLE 1000 1000)
ACTION:(CONCLUDE CNTXT SS-STRESS FATIGUE TALY 1.0)

UPCOMING EXPERT SYSTEMS IN THE NEXT GENERATION?

Advances in computer hardware have contributed significantly to
the expansion of expert systems.
EXPERT SYSTEMS........... INTELLIGENT SYSTEMS:
Computer hardware advances have made possible "integrated
expert Systems", that is, expert systems embedded in microprocessor
chips to form an integrated hardware/software package. These
integrated experts systems can be embedded in a piece of Equipment,
such as Complex electronic gear, to form what we will call an
"intelligent System".

HEIRARCHICAL INTELLIGENT SYSTEMS:

11 Email: chinna_chetan05@yahoo.com
Visit: www.geocities.com/chinna_chetan05/forfriends.html

The intelligent system configuration will be particularly useful
when the equipment to be monitored forms a hierarchy of physical
units arranged in some Network structure.

 An intelligent system has already been developed and put into
commercial use.
 The SPE expert system runs on a microprocessor.
 This expert system interprets waveforms from destinometer to
determine which of several diseases a patient might have.

EXPERT SYSTEM FOR 'THE LAW':

Expert systems are beginning to be noticed by the legal
profession.
Four important types of applications for expert systems in law exist.
They are:

1. Document Generation: Produces legal documents such as wills,
contracts, draft legislation by selecting or composing appropriate
pieces of text selecting or composing appropriate pieces of text and
organizing them into document form.

2. Interpretation and Prediction: Interprets the law in the context
of a particular question or problem, anticipates the legal consequences
of proposed actions and predicts the effects of changes in legislation.

3. Scheduling an Monitoring: Schedules attorneys activities and
periodically inspects legal data bases and knowledge bases for
changes in the law that could affect clients and active cases.

4. Case Management: Organizes case information, estimates case
value, and suggests for negotiation and case disposition.

Common pitfalls & Limitations in expert system
implementation:
PITFALLS:
1. The expert system development effort is addressing a problem
so difficult that it can't be solved within the constraints set by the
available resources.
2. The problem that the expert system is designed to solve will not
significantly alleviates the difficulty that motivated the
development effort.

12 Email: chinna_chetan05@yahoo.com
Visit: www.geocities.com/chinna_chetan05/forfriends.html

3. The problem that the expert system addresses is so general or
complex that an excessive number of rules and database objects
are needed to describe the expertise adequately.

LIMITATIONS:
 Expert systems are not good at representing temporal
knowledge.
 Expert systems are not good at representing Spatial
Knowledge
 Expert systems are not good at performing commonsense.
 Expert systems are not good at recognizing their ability.
 Expert systems are not good at handling the inconsistent
knowledge.

Conclusion:
Human brain is on an everlasting search for something
innovative and so is the case Expert systems are the invention that
was fuelled by the idea of making machines intelligent. The expert
system area is expanding rapidly, Both government and industry are
beginning to invest in commercial expert systems, and with in a few
years a very large number of Indian companies in collaboration with
U.S. companies will be involved in AI or Expert System research and
development.

At last, I conclude that one of the most important by-products of
this expert system development will be the codification of
knowledge. Advances in computer hardware have contributed
significantly to the expansion of Expert Systems. Undoubtedly we can
expect a lot from these in future to say truly there’s no limit for what it
can do.

Bibliography:
1. A Guide to EXPERT SYSTEMS - Donald A. Waterman

2. EXPERT SYSTEMS -Davis. R.

13 Email: chinna_chetan05@yahoo.com
Visit: www.geocities.com/chinna_chetan05/forfriends.html

3. EXPERT SYSTEMS Principles and Programming, 3rd Edition,

-Giarratano & Riley.

4. Artificial Intelligence A New Synthesis. - Nils. Nilsson.

5. Introduction to Artificial Intelligence AND EXPERT SYSTEMS

– Dan. W. Patterson.

14 Email: chinna_chetan05@yahoo.com
Visit: www.geocities.com/chinna_chetan05/forfriends.html

15 Email: chinna_chetan05@yahoo.com