Expert Systems in AI

• An expert system is a computer system that emulates the decision-making capabilities of a human expert
• Professor Edward Feigenbaum of Stanford University defined expert systems

Areas of Artificial Intelligence (AI)

• Artificial Intelligence encompasses Robotics, Speech,Vision, Natural Language, Understanding, Artificial Neural Systems, and Expert Systems

Expert System Technology May Include

• Special expert system languages like CLIPS
• Programs
• Hardware designed to facilitate system implementation

Expert System Main Components

• Knowledge base: obtainable from books, magazines, knowledgeable persons
• Inference engine: draws conclusions from the knowledge base

Problem Domain vs. Knowledge Domain

• An expert's knowledge is specific to a problem domain like medicine, finance, science or engineering
• An expert's knowledge about solving specific problems is the knowledge domain
• The problem domain is always a superset of the knowledge domain

Advantages of Expert Systems

• Increased availability
• Reduced cost and danger
• Better performance
• Multiple expertise
• Increased reliability
• Provides explanations
• Fast response times
• Steady, unemotional, and complete responses at all times
• Functions as an intelligent tutor and database

Representing Knowledge

• Knowledge in an expert system can be represented in multiple ways, including IF-THEN rules

Knowledge Engineering

• Developing an expert system involves a knowledge engineer eliciting knowledge from an expert
• The knowledge engineer codes the knowledge explicitly in the knowledge base
• The expert evaluates the expert system and critiques the knowledge engineer

The Role of AI

• An algorithm is an ideal solution guaranteed to yield a solution in a finite amount of time
• When an algorithm is not available, artificial intelligence(AI) is used instead
• Expert systems use inference to arrive at a "reasonable solution"

Uncertainty

• Both human experts and expert systems must handle uncertainty
• Easier to program expert systems with shallow knowledge than with deep knowledge
• Shallow knowledge is based on empirical and heuristic knowledge
• Deep knowledge is based on the structure, function, and behavior of objects

Limitations of Expert Systems

• Typical expert systems cannot generalize about new situations like people can
• Knowledge acquisition bottleneck results from the time-consuming and labor-intensive task of building an expert system

Early Expert Systems

• DENDRAL: identifies chemical constituents using chemical mass spectroscopy
• MYCIN: medical diagnosis of illness
• DIPMETER: geological data analysis for oil
• PROSPECTOR: geological data analysis for minerals
• XCON/R1: configures computer systems

Broad Classes of Expert Systems

• Configuration: assembles system components properly
• Diagnosis: infers underlying problems from evidence
• Instruction: intelligent teaching capabilities
• Interpretation: explains observed data
• Monitoring: compares observed data to expected data to judge performance
• Planning: devises actions for desired outcomes
• Prognosis: predicts outcomes of situations
• Remedy: prescribes problem treatments
• Control: regulates a process, incorporating interpretation, diagnosis, monitoring, planning, prognosis, and remedies

Problems with Algorithmic Solutions

• Conventional computer programs solve problems algorithmically
• Algorithmic languages include C, Java, and C#
• Classic AI languages include LISP and PROLOG

Considerations for Building Expert Systems

• Determine if the problem can be solved effectively by conventional programming
• Confirm the need and desire for an expert system
• Ensure a willing human expert is available to cooperate
• Verify that the expert can explain the knowledge understandably to the engineer
• Verify if the problem-solving knowledge is mainly heuristic and uncertain

Languages, Shells, and Tools

• Expert system languages are post-third generation
• Procedural languages (e.g., C) focus on techniques to represent data
• Modern languages (e.g., Java) focus on data abstraction
• Expert system languages (e.g., CLIPS) focus on representing knowledge

Elements of an Expert System

• User interface communicates between user and system
• Exploration facility explains reasoning
• Working memory contains a global database of facts used by rules
• Inference engine makes inferences, deciding which rules are satisfied and prioritizing
• Agenda prioritizes rules created by the inference engine, satisfied by facts in working memory
• Knowledge acquisition facility allows users to input knowledge, bypassing explicit coding

Production Rules

• Knowledge base is also called production memory
• Production rules can be expressed in IF-THEN pseudocode format
• In rule-based systems, the inference engine determines which rule antecedents are satisfied by the facts

General Methods of Inferencing

• Forward chaining reasons from facts to conclusions, is best for prognosis, monitoring, and control
• Backward chaining reasons in reverse from a hypothesis, best for diagnosis problems

Production Systems

• Rule-based expert systems are the most popular type
• Knowledge is represented as rules that specify what should/not be concluded
• Forward chaining starts with facts and applies rules to draw conclusions/take actions
• Backward chaining starts with a hypothesis and seeks rules to prove it true
• Forward chaining is primarily data-driven
• Backward chaining is primarily goal-driven

Post Production System

• A mathematical/logical system with rules that change one string of symbols into another
• The basic limitation of the Post Production System is the lack of control mechanism

Markov Algorithm

• Markov Algorithm is an ordered group of productions with applied priorities to an input string
• If the highest priority rule is not applicable, the next applicable rule is applied
• It is an efficient algorithm for systems with many rules

Rete Algorithm

• The Rete Algorithms act like a net holding information
• Rete Algorithms have faster response times and rule firings as compared to IF-THEN rules
• Takes advantage of temporal redundancy and structural similarity, the drawback is is high memory requirements

Procedural Paradigms

• Algorithm - method of solving a problem in a finite number of steps
• Procedural programs are also called sequential programs
• The programmer specifies exactly how a problem solution must be coded

Imperative Programming

• Focuses on modifiable store -- variables and assignments
• During execution, a program transitions from initial to final state through intermediate states
• Provides for top-down design, but is not efficient for expert systems

Nonprocedural Paradigms

• Does not depend on the programmer giving exact details how the program is to be solved
• Declarative programming separates the goal from the method
• Object-oriented programming is partly imperative and partly declarative using objects and methods
• Inheritance derives OOP subclasses from parent classes
• Nonprocedural Paradigms are be considered declarative languages
• Induction-based programming - the program learns by generalizing from a sample

Summary

• During the 20th Century various definitions of AI were proposed
• In the 1960s, a special type of AI called expert systems dealt with complex problems in a narrow domain, e.g., medical disease diagnosis
• Today, expert systems are used in a variety of fields
• Expert systems solve problems for which there are no known algorithms
• Expert systems are knowledge-based and effective for solving real-world problems
• Expert systems are not suited for all applications
• Future advances in expert systems will hinge on new quantum computers in conjunction with computers on the Internet