ch 1 fundamentals of expert systems

Upload: serak-shiferaw

Post on 06-Apr-2018

223 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    1/48

    5/5/12 Lecture One(MAH)

    Expert Systems

    Chapter One

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    2/48

    5/5/12 Lecture One(MAH)

    Fundamentals of Expert Systems

    Introduction

    Expert System

    Was derived fromthe term

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    3/48

    5/5/12 Lecture One(MAH)

    Fundamentals of Expert SystemsHistory of Expert System

    Es is developed by the AI community early mid 1960s

    During this period of AI research is dominated by a believe

    that few laws of reasoning coupled withpowerful computers

    would produce expert or even super human performance.

    Early Examples: GPS by Newell and Simon

    From their logic theory of machine

    Was an attempt to create an intelligent computer

    Predecessor to ES

    Designed to change a certain initial situation in to

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    4/48

    5/5/12 Lecture One(MAH)

    Systems

    History of ..(cont)

    It also has an optional set ofheuristics for operators to tryfirst

    In ES terms these form a rule base

    GPS attempts to find list of operators that reduce thedifference between a goal and current states

    Sometimes, the operators cannot operate on the current states(their preconditions are not suitable)

    GPS sets itself a sub goal to change the current state into onethat is suitable for the operators

    Many such sub goals may have to be set before GPS can

    solve a problem

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    5/48

    5/5/12 Lecture One(MAH)

    Systems

    History of .(cont)

    Early expert system Marked by Shift from general purpose to special purposeprograms

    Mid 1960s with the development of DENDRAL by E. Feigenbaun atStanford university, followed by the development ofMYCIN and when

    researchers also recognized that the problem solving mechanism is onlya small part of a complete, intelligent computer system

    DENDRALconstruction led the following conclusions

    GPS are too weak

    Human problem solverare good only if they operate in a very narrow domain

    Expert systems need to be constantly updated for new information(rule basedrepresentation is needed) and the complexity of problems requires aconsiderable amount of knowledge about the problem area

    Several expert systems had begun to emerge(reading assignment)

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    6/48

    5/5/12 Lecture One(MAH)

    Systems

    History of .(cont)

    But due to similar reasons to the general problem solvers,comprehensive knowledge had limited success

    Knowledge-based problems in general werepremature

    Knowledge as a target of study is too broad and diverse

    Nevertheless, several different approaches to knowledge representationproved sufficient for the expert systems that employed them

    Key insight learned at that time was the power of an ES that can bederived from the knowledge it possesses not from the particular

    formalisms and inference scheme it employs(Expert knowledgeper seseems both necessary and sufficient to develop an expert system)

    Beginning of the 1980s, ES technology, first limited to the academicscene, began to appear as commercial applications XCON (Digital

    Equipment Corp.), XSEL(digital equipment corp.) and CATS (GeneralElectric

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    7/48

    5/5/12 Lecture One(MAH)

    SystemsHistory of.(contd)

    Programming tools (EMYCIN, AGE, EXPERT, KAS)

    Tools for learning from experience(META,DENDRAL, EURISKO)

    Commercially available starting in 1983

    Most of the development tools required special hardware(LISP

    machines)

    But the late 1980s, development software can run on regular computersincluding microcomputers

    Latest developments of in Expert system area

    Availability ofmany tools that are designed to expedite the construction of ESat a reduced cost

    Dissemination of ES in thousands of organizations, some of which has manyspecific systems

    Increased use of expert system in many tasks

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    8/48

    5/5/12 Lecture One(MAH)

    un amen a s o xperSystemsHistory of(contd)

    Use of ES technology as a methodology for expediting the constructionof regularinformation systems

    Increased use of the object-oriented programming approach inknowledge representation

    Development ofcomplex systems with multiple sources of knowledge,multiple lines of reasoning and fuzzy information

    Use of multiple knowledge base

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    9/48

    5/5/12 Lecture One(MAH)

    Historical Overview: Detail

    At the beginning(1956) all projects use the same programminglanguage such as LISP orPROLOG.

    List processing Language

    PROgramming in LOGic

    LISP

    It created in the late 1950s.

    Facilitate symbol manipulation (Intelligent behaviors can berepresented easily)

    It was found to be difficult to master.

    So that its use is limited to AI research programs in academic

    circle.

    Fundamentals of ExpertSystems

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    10/48

    5/5/12 Lecture One(MAH)

    Historical Overview: Detail LISP

    In addition, the many dialects of LISP is a problem.

    Fortunately, this situation improved in mid-1970s with the

    introduction of a LISP standard called Common LISP. LISP has its roots in one area of mathematics (lambda calculus) .

    Fundamentals of ExpertSystems

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    11/48

    5/5/12 Lecture One(MAH)

    Historical Overview: Detail

    PROLOG In the early 1970s PROLOG invented in France.

    It also roots in one area of mathematics (first-order predicatecalculus).

    PROLOG, however, did not immediately become a language ofchoice for AI programmers .

    Until the early 1980s when the Japanese use a logic programminglanguage for the Fifth Generation Computing Systems (FGCS)

    Project. After than researchers in the U.K. and Japan adopted PROLOG for

    developing intelligent programs.

    It consists of English-like statements which are facts, rules, andquestions.

    Both LISP PROLOG re uired a disci lined student to master it

    Fundamentals of ExpertSystems

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    12/48

    5/5/12 Lecture One(MAH)

    Historical Overview: Detail

    PROLOG Due to this until the years of1970s few expert systems were built.

    Also, since these systems were built from scratch, development timewas large.

    MYCIN project that dramatically changed this situation. (1976,Stanford University to aid physicians)

    To diagnosing and treating patients with infectious blooddiseases caused by bacteremia .

    MYCIN

    Took approximately 20 person-years to complete.

    MYCIN is a rule-based expert system.

    That usesbackward chaining and

    Fundamentals of ExpertSystems

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    13/48

    5/5/12 Lecture One(MAH)

    HistoricalOverview:Detail

    MYCIN At the end of the project, the MYCIN developers realized that by

    separating the knowledge on infectious diseases from its control,then the code written for the other modules should be portable toother applications.

    By removing the knowledge about infectious blood diseases, asystem known as EMYCIN was formed.

    EMYCIN facilitated the development of other expert systems, suchas PUFF an application for the diagnosis ofpulmonary problems.

    The separation of knowledge from its processing is a powerfulfeature of expert systems that permits the reuse of existing code andgreatly reduces the development time for other systems.

    Fundamentals of ExpertSystems

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    14/48

    5/5/12 Lecture One(MAH)

    Historical Overview: Detail For example, PUFF was produced in about 5 person-years.

    when the number of shell vendors began to grow rapidly.

    Generally the dominant languages used for building an expert systemhave been LISP, PROLOG and OPS (Official Production System ).

    Recently, C and C++ have also been used for system development.Software used inexpert systemdevelopment.

    Fundamentals of ExpertSystems

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    15/48

    5/5/12 Lecture One(MAH)

    Historical Overview: Detail

    OPS provides flexibility by providing many of the needed facilitiessuch as the inferenceengine and explanation facility.

    The percentage of systems built using shells has increased slightly overthe 1993, while percentages for the languages have all decreased

    slightly.

    Number of developed expert systems peryear

    Fundamentals of ExpertSystems

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    16/48

    5/5/12 Lecture One(MAH)

    SystemsBasic Concepts of Expert systems which are the core

    concepts of the course The CATS-1 example introduces the basic concepts

    of expert systems:

    Expertise

    The extensive, task-specific knowledgeacquired from training, reading, and

    experience that enable experts to make better

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    17/48

    5/5/12 Lecture One(MAH)

    Systems Rules(heuristics) of what to do in a given problem situation( rules regarding

    problem solving)

    Global strategies for solving problems Meta-knowledge (knowledge about knowledge)

    It takes a long time (usually several years) to become an expert, and novices become experts only incrementally

    Experts:

    Difficult to define because of levels of degrees or level of expertise (how much expertise should a person

    possess before qualifying as an expert)

    Non experts outnumber experts in many fields

    It is possible to increase top level expertise available to other decision makers

    Human expertise includes a constellation of behavior that involves the following activities that must

    be done efficiently (quickly and at low cost) and effectively(a high quality result)

    Recognizing and formulating the problem

    Solving the problem quickly and properly( ES primarily employed)

    Explaining the solution (ES primarily Employed)

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    18/48

    5/5/12 Lecture One(MAH)

    Systems To mimic the human expert, it is necessary to build a computer thatexhibits all the above characteristics

    Transferring expertise:

    the objective of an expert system is to transfer expertise from an expert

    to a computer and then on to other humans(non experts).

    It involves four activities:

    Knowledge acquisition from experts or other sources Knowledge representation in the computer

    Knowledge inferencing

    Knowledge transfer to the user

    The knowledge is stored in the computer in a component called aknowledge base. Two types of knowledge are distinguished: facts and

    procedures(usually rules) regarding the position domain

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    19/48

    5/5/12 Lecture One(MAH)

    SystemsInferencing:

    A unique feature of an expert system is its ability to reason

    The attempt(aim) is that all the expertise is stored in the knowledgebase and that the program has accessibility to databases, the computeris programmed so that it can make inference

    The inferencing is performed in a component called the inferenceengine, which includes procedure regarding problem solving

    Rules:

    Most commercial ES are rule based systems; that is, the knowledge isstored mainly in the form of rules, as are the problem solving procedures.

    A rule in the CATS-1 example may look like this:If the engine is idle,

    and the fuel pressure is less than 38 psi, and the gauge is accurate, thenthere is a fuel system fault(60 rules of such type are there in CATS-1system

    Frame representation is complementing the rule representation

    n amen a s o per

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    20/48

    5/5/12 Lecture One(MAH)

    un amen a s o xperSystems

    Explanation capability:

    Another unique feature of ES

    It is expert systems ability to explain its advice or recommendationsand even to justify why a certain action was not recommended

    the explanation and justification is done in a sub system called justifieror explanation system to examine its own reasoning and to explain itsoperation

    The characteristics and capabilities of ES make them different fromconventional systems

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    21/48

    5/5/12 Lecture One(MAH)

    SystemsConventional Systems Expert SystemsInformation and its processing are usually combined intoone sequential program

    Knowledge base is clearly separated from theprocessing(inference) mechanism (knowledge rulesseparated from control)

    Program does not make mistakes (programmers do)Program may make mistakes

    Do not(usually) explain why input data are neededorhow conclusions are drawn

    Explanation is a part of most ES

    Changes in the program are tedious Changes in the rules are easy to accomplish

    The system operates only when it is completed The system can operate with only a few rules

    Execution is done on a step-by-step(algorithmic) basis Execution is done by using heuristics and logic

    Need complete information to operate Can operate with incomplete or uncertain information

    Effective manipulation of large databases Effective manipulation oflarge knowledge bases

    Representation and use ofdata Representation and use of knowledge

    Efficiency is a major goal Effectiveness is the major goal

    Easily deal with quantitative data Easily deal with qualitative data

    Capture, magnify, and distribute access to numeric dataor to information

    Capture, magnify, and distribute access tojudgment anknowledge

    F d t l f E t S t

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    22/48

    5/5/12 Lecture One(MAH)

    Fundamentals of Expert Systems

    Structure of Expert System

    Expert systems are composed of two major parts: the development andthe consultation(runtime).

    The development environment is used by ES builder to build thecomponents and to introduce a non expert to obtain expert knowledge

    and advice. The following components may exist in an expert system:

    Knowledge acquisition subsystem:

    is the accumulation, transfer, and transformation of problemsolving expertise from some knowledge source to a computerprogram for constructing or expediting the knowledge base.

    Potential sources of knowledge include human experts, textbooks, databases, special research reports, and pictures

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    23/48

    5/5/12 Lecture One(MAH)

    Fundamentals of Expert Systems

    Structure of Expert .(contd)

    Knowledge acquisition systemcontd

    Acquiring knowledge from experts is a complex task thatfrequently creates a bottleneck in ES construction.

    The state of the art today requires a knowledge engineer tointeract with one or more human experts in building theknowledge base

    Typically, the knowledge engineer helps the expert structure the

    problem area by interpreting and integrating human answers toquestions, drawing, analogies, posing counter examples, and

    bringing to light conceptual difficulties

    Knowledge base:

    Contains necessary elements for understanding, formulating and

    F d t l f E i t

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    24/48

    5/5/12 Lecture One(MAH)

    Fundamentals of ExperimentStructure of Expert .(contd)

    Knowledge Base.contd Includes two basic elements:

    Facts, such as the problem situation and theory of the problem area

    Special heuristics, or rules that direct the use of knowledge to solvespecific problems in a particular domain. The heuristics express theinformal judgmental knowledge in an application area.

    Global strategies , which can be both heuristics and a part of atheory of the problem area, are usually included in the

    knowledge base. Knowledge, not mere facts, is the primarymaterial of expert systems.

    The information in the knowledge base is incorporated in thecomputer program by a process called knowledge representation

    Fun amenta Exper ment

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    25/48

    5/5/12 Lecture One(MAH)

    Fun amenta Exper mentStructure of Expert .(contd)

    Inference Engine

    The brain of the ES is the inference engine, also known as thecontrol structure orthe rule interpreter(in rule-based ES)

    This component is essentially a computer program that provides amethodology for reasoning about information in a knowledge baseand in the blackboard, and forformulating conclusions

    This component provides directions about how to use the systemsknowledge by developing the agenda that organizes and controls thesteps taken to solve problems whenever consultation is performed

    It has three major elements:

    An interpreter(rule interpreter in most systems) , which executesthe choice agenda items by applying the corresponding knowledge

    base rules)

    Fundamental Experiment

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    26/48

    5/5/12 Lecture One(MAH)

    Fundamental Experiment

    A scheduler, which maintains control over theagenda. It estimates effects of applying inferencerules in light ofitem priorities or other criteria on theagenda

    A consistency enforcer, which attempts to maintain aconsistent representation of the emerging solution

    Blackboard:

    Is an area of working memory set aside for the descriptionof a current problem, as specified by the input data

    It is also used for recording intermediate results

    It records intermediate hypotheses and decisions

    Inference Engine

    Structure of Expert .(contd)

    Fundamental experiment

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    27/48

    5/5/12 Lecture One(MAH)

    Fundamental experiment

    Three types of decisions can be recorded on the blackboard Plan-how to attack the problem

    Agenda-potential actions awaiting execution

    Solution-candidate hypotheses and alternative courses of action thatthe system has generated thus far

    It exists only in some systems

    User Interface:

    Expert systems contain a language processor for friendly problem-oriented communication between the user and the computer.

    This communication could best be carried out in natural language, and insome cases it is supplementedby menus and graphics

    Black Board

    Structure of Expert .(contd)

    Fundamentals of Expert System

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    28/48

    5/5/12 Lecture One(MAH)

    Fundamentals of Expert SystemStructure of Expert .(contd)

    Explanation subsystem(justifier) The ability to trace responsibility for conclusions to their

    sources is crucial both in the transfer of expertise and inproblem solving

    The explanation subsystem can trace such responsibilityand explain the ES behavior by interactively answeringquestions such as the following:

    Why was a certain question asked by the expert system? How was a certain conclusion reached?

    Why was a certain alternative rejected?

    What is the plan to reach the solution? e.g. what remains to beestablished before a final diagnosis can be determined?

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    29/48

    5/5/12 Lecture One(MAH)

    System

    Structure of Expert .(contd)

    Knowledge Refining System:

    Human experts have a knowledge refining system; that is, they cananalyze their own performance, learn from it, and improve it for future

    consultations.

    Similarly, such evaluation is necessary in computerized learning so thatthe program will be able to analyze the reasons for its success orfailure. This could lead to improvements that result in a better

    knowledge base and more effective reasoning

    This component is not available in commercial expert systems, butavailable in experimental expert systems in academics

    E i t

    Development

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    30/48

    5/5/12 Lecture One(MAH)

    Environment

    Blackboard(workplace)Plan Agenda

    Solution ProblemDescri tion

    UserInterface

    Explanation

    Facility

    RecommendedAction

    KnowledgebaseFacts what is known about

    the domain areaRules Logical reference(e.g.Between symptoms and

    causes)

    Inference EngineInterpreterDraws Conclusions

    Scheduler ConsistencyEnforcer Knowle

    dgeRefine

    ment

    ExpertKnowled

    ge

    Knowledge

    Engineer

    ExpertKnowled

    ge

    Facts aboutthe specific

    incident

    KnowledgeAcquisiti

    on

    DevelopmentEnvironment

    Fundamentals of Experiment System

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    31/48

    5/5/12 Lecture One(MAH)

    Fundamentals of Experiment System

    The human element in expert systems

    At least two humans or more participate in the development and

    use of an expert system At a minimum, there is an expert and a user

    Frequently, there is also a knowledge engineer and a systembuilder

    (Domain)Expert: A person who has the special knowledge, judgment,

    experience, and methods along with the ability to apply thesetalents to give advice and solve problems

    It is the domain experts job to provide knowledge about howhe or she performs the task that the knowledge system will

    perform

    The expert knows which facts are important and understandsthe meaning of the relationships among facts

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    32/48

    5/5/12 Lecture One(MAH)

    System

    The human element of expert system

    Domain expert ..

    The usual initial body of knowledge , including terms andbasic concepts , is documented in textbooks, reference

    manuals, sets of policies, or a catalogue of products However, this is not sufficient for powerful ES

    Not all expertise can be documented because most experts areunaware of the exact mental process by which they diagnoseor solve problem

    Thus, an interactive procedure is needed to acquire additionalinformation from the expert to expand the basic knowledge

    The process is fairly complex and usually requires the

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    33/48

    5/5/12 Lecture One(MAH)

    System

    The human element in expert system

    The Knowledge Engineer

    The knowledge engineer helps the expert(s) in structuring the problemarea by:

    Interpreting and integrating to questions by

    Drawing analogies posing counterexamples

    Bringing to light conceptual difficulties

    Knowledge engineer is also a system builder

    Shortage of knowledge engineers is a major bottleneck in ESconstruction

    To overcome this bottleneck ES designers are using productivitytools(special editors)

    Research is being conducted on building systems that will bypass the

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    34/48

    5/5/12 Lecture One(MAH)

    System

    The human element in expert system

    The user Most computer-based systems have evolved in a single-user mode

    In contrast an ES has several possible types of users:

    A non-expert client seeking direct advice to act as a consultant oradvisor

    A student who want to learn so that ES can act as instructor

    An ES builder who wants to improve or increase the knowledge

    base so that ES can act as a partner (collaborator) An expert so that ES acts as a colleague(assistant)

    The knowledge engineer and the domain expert should anticipate usersneeds and limitations when designing ES

    The capabilities of ES were developed to save users time and effort

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    35/48

    5/5/12 Lecture One(MAH)

    Systems

    Other participants

    A System builder helps to integrate the expert system withother computerized systems

    A tool builder provide generic or build specific tools

    Vendors that provide tools and advice Support staff provide clerical and technical help

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    36/48

    5/5/12 Lecture One(MAH)

    Participants of ES

    Tools,language

    Vender

    Expertsystem

    Knowledge

    engineer

    Systembuilder

    Expert

    Documented

    knowledge

    Supportstaff

    End user

    Toolbuilder

    Provide

    Test

    Us

    e

    Bui

    ld

    Acquir

    ingKnowledge

    Use Build

    Build Conn

    ect

    Use

    Sup

    porttasks

    Corporate

    Fundamentals of expertSystems

    p

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    37/48

    5/5/12 Lecture One(MAH)

    FundamentalsofExpert

    How expert system work

    Development: construction of knowledge base(fact and

    procedures), inference engine(development and

    acquisition), blackboard, explanation facility, interfaces)

    Consultation: Transferring to users and consulting it when

    they need advice by conducting bidirectional dialoguewith the system so that they can get solutions in terms of

    conclusions

    Improvement of ES several times through a process called

    rapid prototyping during their development

    Fundamentals of Expert

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    38/48

    5/5/12 Lecture One(MAH)

    Process of building (developing) ES can be lengthy.

    A tool that is frequently used to expedite

    development called ES shell is used

    ES shell include all the generic components of an

    ES but they do not include the knowledge. E.g. :

    EMYCIN

    Problem Areas Addressed by Expert Systems

    ES systems can be classified in several ways. One

    way of classifying is a generic categorization that

    uses the general problem areas they address.

    Fundamentals of Expertsystem

    S t

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    39/48

    5/5/12 Lecture One(MAH)

    Systems Generic Categories of Expert Systems

    Category Problem addressed

    Interpretation Inferring situation descriptions from observations (includes surveillance, speech

    understanding, image analysis, signal interpretation, and many kinds of intelligenceanalysis)

    Prediction Inferring likely consequences of given situations (weather forecasting, demographicpredictions, economic forecasting, traffic predictions, crop estimates, and military,marketing or financial forecasting)

    Diagnosis Inferring system malfunctions from observations( medical, electronic, mechanical,and software)

    Design Configuring objects under constraints (circuit layout, building design, and plantlayout)

    Planning Developing plans to achieve goal(s) (routing, communications, productiondevelopment, etc)

    Monitoring Comparing observations to plans(standards), flagging exceptions(air traffic, etc)

    Debugging Prescribing remedies for malfunctions(

    Repair Executing a plan to administer a prescribed remedy

    Instruction Diagnosing, debugging, and correcting student performance

    Control Interpreting, predicting, repairing, and monitoring system behaviors

    S t

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    40/48

    5/5/12 Lecture One(MAH)

    Tasks suitable for expert systems

    To easy (Requiresconventional software)

    Just right Too hard (requires humanintelligence)

    Payroll, inventory Diagnosing andtroubleshooting

    Designing new tools or acover for magazine

    Sample tax returns Analyzing diverse data Stock market prediction

    Decision trees Production scheduling Discovering new principles

    Database management Equipment layout Every daylanguage(commonsense)

    problems

    Mortgage computation Advise on tax shelters Developing new statisticaltests

    Regression analysis Determine type of

    statistical analysis

    Require innovation or

    discovery(commonsense)

    Systems

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    41/48

    5/5/12 Lecture One(MAH)

    Benefits of Expert systems

    Increased output and productivity: Es can work faster than

    humans Increased Quality: ES can increase quality by providing consistent

    advice and reducing error rate

    Reduce downtime: many operational ES are used for diagnosing

    malfunctions and prescribing repairs Capture of scarce resources(leave, retire, needed across a broad

    geography)

    flexibility in providing services and in manufacturing

    Easier equipment operation: Es makes complex equipment easierto operate. E.g. STEAMER is an ES intended to traininexperienced workers to operate complex ship engines

    Systems

    F d t l f t S t

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    42/48

    5/5/12 Lecture One(MAH)

    Fundamentals of expert Systems Benefits of .

    Increased capabilities of other computerized systems(Integration) Integration several expert opinions that increase the quality of advice

    Ability with to work with incomplete or uncertain information . A usercan interact with the system with do not know or not sure answer

    Provision of training: ES can provide training

    Enhancement of problem solving as it allows the integration of topexperts judgment into analysis. They increase users understandingthrough explanation

    Ability to solve complex problems but in narrow domain

    Knowledge transfer to remote locations which is more important todeveloping countries

    Fundamentals of expert Systems

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    43/48

    5/5/12 Lecture One(MAH)

    Benefits of .

    Elimination of the need for expensiveequipment because of their ability to control

    thoroughly and quickly the information

    provided by instruments

    Operation in hazardous environments :e.g.

    in military conflicts, hot, humid, or toxicenvironments

    Accessibility to knowledge and helpdesks:

    Fundamentals of expert Systems

    Fundamentals of expert Systems

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    44/48

    5/5/12 Lecture One(MAH)

    Problem and limitation of expert systems

    Available Es methodologies are not straight forward and effective, even

    for application in generic categories. Some ES codes, especially forsystems constructed with programming languages , is generally hard tounderstand

    Knowledge is not readily available

    Expertise is hard to extract from humans

    The approach to each expert to situation assessment may be different,yet correct

    It is hard even for highly skilled professional to abstract goodsituational

    Assessments when he or she is under time pressure

    Users of expert systems have natural cognitive limits

    ES work well only in a narrow domain, in some cases in a very narrow

    Fundamentals of expert Systems

    un amen a s o exper ys ems

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    45/48

    5/5/12 Lecture One(MAH)

    The vocabulary or jargon

    that experts use for

    expressing facts and

    relations is frequently

    limited and not understoodby others

    un amen a s o exper ys ems

    Representative tasks of experts and their

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    46/48

    5/5/12 Lecture One(MAH)

    Representative tasks of experts and theirdifficulties

    Task Difficulties

    Interpretation: analysis of data to determinetheir meaning

    Data are often noisy and fullof errorsData value may be missing

    Diagnosis: faultfinding in a system based oninterpretation of data

    Faults can be intermittentSymptoms of the faults mayinterfereData contain errors or areinaccessibleDiagnostic equipment may be

    unreliable

    Monitoring: continuously interpretingsignals and flag for intervention

    When to flag often depends oncontextSignal expectation vary with thetime/situation

    Prediction: forecasting from past andpresent

    Integration of incompleteinformationAccount for multiple possiblefeaturesContingencies for uncertaintiesDiversity of data, often

    contradicting data

    Fundamentals of expert Systems

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    47/48

    5/5/12 Lecture One(MAH)

    Types of Expert Systems

    Knowledge based systems can be built more easily

    and quickly than expert systems. What distinguishes

    expert systems from knowledge based systems is

    their the amount of expertise they have.

    Types

    Rule-based expert systems: are mostly

    commercial. The technology is relatively well

    developed( E.g. MYCIN)

    Frame based expert systems: Knowledge is

    represented as frames, representation of the OOP

    Fundamentals of expert Systems

    Fundamentals of expert Systems

  • 8/3/2019 Ch 1 Fundamentals of Expert Systems

    48/48

    Ready made(turnkey) systems developed to satisfy a

    particular needs of a users(custom made) or they canbe purchased as ready-made packages for any users.

    Ready made systems are similar to application

    packages. They are considered as less expensive

    than customized systems.

    But they are general in nature and their value maynot important to the user. are not popular.

    Real-time expert systems: systems in which there is

    Fundamentals of expert Systems