Chapter 1: Introduction To Expert Systems - WordPress

Chapter 1:Introduction toExpert SystemsExpert Systems: Principles andProgramming, Fourth EditionOriginal by Course TechnologyModified by Ramin Halavati (halavati@ce.sharif.edu)

Objectives Learn the meaning of an expert system Understand the problem domain and knowledgedomain Learn the advantages of an expert system Understand the stages in the development of anexpert system Examine the general characteristics of an expertsystemExpert Systems: Principles and Programming, Fourth Edition2

Objectives Examine earlier expert systems which have givenrise to today’s knowledge-based systems Explore the applications of expert systems in usetoday Examine the structure of a rule-based expertsystem Learn the difference between procedural andnonprocedural paradigms What are the characteristics of artificial neuralsystemsExpert Systems: Principles and Programming, Fourth Edition3

What is an expert system?“An expert system is a computer system thatemulates, or acts in all respects, with thedecision-making capabilities of a human expert.”Professor Edward FeigenbaumStanford UniversityExpert Systems: Principles and Programming, Fourth Edition4

Fig 1.1 Areas of ArtificialIntelligenceExpert Systems: Principles and Programming, Fourth Edition5

Expert system technologymay include: Special expert system languages – CLIPS Programs Hardware designed to facilitate theimplementation of those systemsExpert Systems: Principles and Programming, Fourth Edition6

Expert System Main Components Knowledge base – obtainable from books,magazines, knowledgeable persons, etc. Inference engine – draws conclusions from theknowledge baseExpert Systems: Principles and Programming, Fourth Edition7

Figure 1.2 Basic Functionsof Expert SystemsExpert Systems: Principles and Programming, Fourth Edition8

Problem Domain vs. KnowledgeDomain An expert’s knowledge is specific to one problemdomain – medicine, finance, science,engineering, etc. The expert’s knowledge about solving specificproblems is called the knowledge domain. The problem domain is always a superset of theknowledge domain.Expert Systems: Principles and Programming, Fourth Edition9

Figure 1.3 Problem andKnowledge Domain RelationshipExpert Systems: Principles and Programming, Fourth Edition10

Advantages of Expert Systems Increased availability Reduced cost Reduced danger Performance Multiple expertise Increased reliabilityExpert Systems: Principles and Programming, Fourth Edition11

Advantages Continued Explanation Fast response Steady, unemotional, and complete responses atall times Intelligent tutor Intelligent databaseExpert Systems: Principles and Programming, Fourth Edition12

Representing the KnowledgeThe knowledge of an expert system can berepresented in a number of ways, including IFTHEN rules:IF you are hungry THEN eatExpert Systems: Principles and Programming, Fourth Edition13

Knowledge EngineeringThe process of building an expert system:1. The knowledge engineer establishes a dialogwith the human expert to elicit knowledge.2. The knowledge engineer codes the knowledgeexplicitly in the knowledge base.3. The expert evaluates the expert system andgives a critique to the knowledge engineer.Expert Systems: Principles and Programming, Fourth Edition14

Development of an Expert SystemExpert Systems: Principles and Programming, Fourth Edition15

The Role of AI An algorithm is an ideal solution guaranteed toyield a solution in a finite amount of time. When an algorithm is not available or isinsufficient, we rely on artificial intelligence(AI). Expert system relies on inference – we accept a“reasonable solution.”Expert Systems: Principles and Programming, Fourth Edition16

Uncertainty Both human experts and expert systems must beable to deal with uncertainty. It is easier to program expert systems withshallow knowledge than with deep knowledge. Shallow knowledge – based on empirical andheuristic knowledge. Deep knowledge – based on basic structure,function, and behavior of objects.Expert Systems: Principles and Programming, Fourth Edition17

Limitations of Expert Systems Typical expert systems cannot generalize throughanalogy to reason about new situations in the waypeople can. A knowledge acquisition bottleneck results fromthe time-consuming and labor intensive task ofbuilding an expert system.Expert Systems: Principles and Programming, Fourth Edition18

Development of Expert Systems Rooted from Cognitive Studies:– How does human process information Newell/Simon Model (GPS)– Long Term Memory: IF-Then Rules– Short Term Memory: Current Facts– Inference Engine/Conflict ResolutionExpert Systems: Principles and Programming, Fourth Edition19

Rule Examples IF the car doesn’t run and the fuel gaugereads empty THEN fill the gas tank. IF there is flame, THEN there is a fire. IF there is smoke, THEN there may be a fire. IF there is a siren, THEN there may be a fire.Expert Systems: Principles and Programming, Fourth Edition20

Expert Knowledge Base Knowledge / Expert Knowledge– Book Rules / Heuristics and Experiences(secrets!) Experts usually score almost similar to novices inbrand new problems.– Chess Rules / Chess Master PatternsExpert Systems: Principles and Programming, Fourth Edition21

Early Expert Systems DENDRAL – used in chemical massspectroscopy to identify chemical constituents MYCIN – medical diagnosis of illness DIPMETER – geological data analysis for oil PROSPECTOR – geological data analysis forminerals XCON/R1 – configuring computer systemsExpert Systems: Principles and Programming, Fourth Edition22

Expert SystemsApplications and DomainsExpert Systems: Principles and Programming, Fourth Edition23

Considerations for BuildingExpert Systems Can the problem be solved effectively by conventionalprogramming?– Ill-Structured Problems / Rigid Control Is the domain well bound?– Headache: Neurochemistry, biochemistry, chemistry, molecularbiology, physics, yoga, exercise, stress management, psychiatry, Is there a need and a desire for an expert system?– The Traffic Light ExampleExpert Systems: Principles and Programming, Fourth Edition24

Considerations for BuildingExpert Systems Is there at least one human expert who is willing tocooperate?– Their faults may b revealed.– Their secrets are revealed.– They have different ideas. Can the expert explain the knowledge to the knowledgeengineer can understand it.– How do you move your finger?– Medicine Is the problem-solving knowledge mainly heuristic anduncertain?– If not, why expert system?Expert Systems: Principles and Programming, Fourth Edition25

Expert SystemsLanguages, Shells, and Tools Conventional computer programs generally solveproblems having algorithmic solutions. Tight interweaving of data and knowledge resultsin rigid control flow control. More advance languages limit the usage, but areeasier for the limited area.Expert Systems: Principles and Programming, Fourth Edition26

Languages, Shells, and Tools Expert system languages are post-third generation. Procedural languages (e.g., C) focus on techniques torepresent data. More modern languages (e.g., Java) focus on dataabstraction. Expert system languages (e.g. CLIPS) focus on ways torepresent knowledge.Expert Systems: Principles and Programming, Fourth Edition27

Elements of an Expert System User interface – mechanism by which user and systemcommunicate. Exploration facility – explains reasoning of expert systemto user. Working memory – global database of facts used by rules. Inference engine – makes inferences deciding which rulesare satisfied and prioritizing.Expert Systems: Principles and Programming, Fourth Edition28

Elements Continued Agenda – a prioritized list of rules created by theinference engine, whose patterns are satisfied by facts orobjects in working memory. Knowledge acquisition facility – automatic way for theuser to enter knowledge in the system bypassing theexplicit coding by knowledge engineer. Knowledge Base!Expert Systems: Principles and Programming, Fourth Edition29

Production Rules Knowledge base is also called productionmemory. Production rules can be expressed in IF-THENpseudocode format. In rule-based systems, the inference enginedetermines which rule antecedents are satisfiedby the facts.Expert Systems: Principles and Programming, Fourth Edition30

An Example from MYCIN IF– The site of the culture is blood and– The identity of the organism is not known withcertainty, and– The stain of the organism is gramnegm and– The morphology of the organism is rod, and– The patient is seriously burned. THEN– There is a weakly suggestive evidence (.4) thatthe identity of the organism is pesudomonas.Expert Systems: Principles and Programming, Fourth Edition31

An Example from XCON/R1 IF– The current context is assigning devices to Unibusmodules, and– There is an unassigned dual-port disk drive, and– The type of controller it requires is known, and– There are two such controllers, neither of which hasany devices assigned to it, and– The number of devices that these controllers cansupport is known, THEN– Assign the disk drive to each of the controllers, and– Note that the two controllers have been associated andeach supports one drive.Expert Systems: Principles and Programming, Fourth Edition32

Structure of aRule-Based Expert SystemExpert Systems: Principles and Programming, Fourth Edition33

General Methods of Inferencing Forward chaining – reasoning from facts to theconclusions resulting from those facts – best forprognosis, monitoring, and control.– primarily data-driven Backward chaining – reasoning in reverse from ahypothesis, a potential conclusion to be proved tothe facts that support the hypothesis – best fordiagnosis problems.– primarily goal drivenExpert Systems: Principles and Programming, Fourth Edition34

Main Inference Engine Cycle While Not DONE– If there are active rules, Conflict Resolution.Else DONE.– Act– Match– Check for Halt End of While Accept a new user command.Expert Systems: Principles and Programming, Fourth Edition35

Mathematical Roots of Rule BasedSystems Post Production Systems Markov Algorithm Rete AlgorithmExpert Systems: Principles and Programming, Fourth Edition36

Post Production System Basic idea – any mathematical / logical system issimply a set of rules specifying how to changeone string of symbols into another string ofsymbols. Basic limitation – lack of control mechanism toguide the application of the rules.Expert Systems: Principles and Programming, Fourth Edition37

Markov Algorithm An ordered group of productions applied in orderor priority to an input string. If the highest priority rule is not applicable, weapply the next, and so on. An efficient algorithm for systems with manyrules.Expert Systems: Principles and Programming, Fourth Edition38

Rete Algorithm Functions like a net – holding a lot of information. Much faster response times and rule firings can occurcompared to a large group of IF-THEN rules whichwould have to be checked one-by-one in conventionalprogram. Takes advantage of temporal redundancy and structuralsimilarity. Drawback is high memory space requirements.Expert Systems: Principles and Programming, Fourth Edition39

Programming Paradigms Procedural (sequential)– Functional/Imperative None ProceduralExpert Systems: Principles and Programming, Fourth Edition40

Procedural Paradigms Algorithm – method of solving a problem in afinite number of steps. Procedural programs are also called sequentialprograms. The programmer specifies exactly how a problemsolution must be coded.Expert Systems: Principles and Programming, Fourth Edition41

Imperative Programming Focuses on the concept of modifiable store – variablesand assignments. During execution, program makes transition from theinitial state to the final state by passing through series ofintermediate states. Provide for top-down-design. Not efficient for directly implementing expert systems.Expert Systems: Principles and Programming, Fourth Edition42