2. 11- 2
Chapter Objectives
1. Define the field of artificial intelligence (AI)
2. Define an expert system (ES).
3. Specify and discuss the areas of ES application.
4. Specify the components of an expert system.
5. Define knowledge base and knowledge representation.
6. Explain what rule-based expert systems are.
7. Define fuzzy logic.
8. Specify the categories of expert system technology.
3. 11- 3
Chapter Objectives, cont.
9. Define the roles in expert system development.
10. Specify the principal benefits and limitations of expert systems.
11. Name other applied fields of artificial intelligence and discuss
their potential role in information systems.
12. Define neural networks and their capabilities.
4. 11- 4
Defining Artificial Intelligence
• AI deals with methods of developing systems that
display aspects of intelligent behavior
• AI systems imitate human capabilities of thinking
and sensing
5. 11- 5
Defining Artificial Intelligence: AI Systems
1. Symbolic Processing
– Computers process symbols
– AI applications process strings of characters that represent the
real world
– Symbols are arranged as lists, hierarchies, or networks and
their interrelations
2. Nonalgorithmic Processing
– Specified step by step procedures
6. 11- 6
Defining Artificial Intelligence
• Science and Technology
• Computer Science
• Biology
• Psychology
• Linguistics
• Mathematics
• Engineering
• Goal: Develop computers that think (reasoning,
learning, and problem solving), sense (see, hear,
talk, feel), and walk
7. 11- 7
Defining Artificial Intelligence:
History and Evolution of AI
• 1950- Turing Test- General problem solving test
• 1960- AI as a field- Knowledge based expert
systems
• 1970- AI commercialization- Transaction
processing and decision support systems
• 1980- Artificial neural networks- Resembling the
human brain
• 1990- Intelligent Agents- Software that performs
assigned tasks
8. 11- 8
Capabilities of Expert Systems: General View
• Expert System (ES)
– Knowledge based system
– Uses inferencing or reasoning procedure to solve problems
that require human expertise
• Knowledge Base
– Domain of knowledge of the expert system
• Heuristic Knowledge
– Rules used by humans
9. 11- 9
Applications of Expert Systems:
Generic Categories of Expert System Applications
• Classification
– Identify an object
• Diagnosis Systems
– Infer malfunction from observable data
• Monitoring
– Continually observe behavior
• Process Control
– Control a physical process based on monitoring
• Design
– Configure a system according to specifications
• Scheduling and Planning
– Plan of action
• Generation of Options
– Alternative solutions to problems
10. 11- 10
How Expert Systems Work
• Knowledge Base
– Organized collection of facts and heuristics about the systems
domain
• Knowledge Representation
– Method to organize the knowledge base
11. 11- 11
Structure of an Expert System
Consultation Environment Development Environment
(Use) (Knowledge Acquisition)
User Expert
Facts of Recommendation,
the Case Explanation
User Interface
Explanation Knowledge
Facility Engineer
Inference Engine
Facts of
the
Knowledge
Case Acquisition
Facility
Working Memory
Knowledge Domain Knowledge
Base (Elements of
Knowledge Base)
12. 11- 12
How Expert Systems Work:
Knowledge Representation
• Frame Based Systems
– Build powerful expert systems
– The frame specifies the attributes of a complex object and its
relationships
• Production Rules
– Rule Based expert systems
– Knowledge is represented by production rules
– Most common method of knowledge representation
– IF part (Condition or Premise) and THEN part (Action or
Conclusions_
– Explanation facility
» How the system arrived at the recommendation
» Uses natural language or numbers
13. 11- 13
How Expert Systems Work:
Inference Engine
• Combines the facts of a specific case with the
knowledge in the knowledge base to decide upon
a recommendation
– Reasoning in Rule Based systems
» In rule based expert system, the inference engine controls
the order in which the production rules are applied or
“fired” and resolves conflicts for more than one applicable
rule
• Directs the user interface to query the user for
further information
– The facts are entered into working memory
– Rules are applied by the inference engine until a goal state is
produced or confirmed
14. 11- 14
How Expert Systems Work:
Inference Engine: Strategies
• Forward Chaining
– A data driven strategy
– Inference from the facts of a case to a conclusion
– Match the IF part with the facts available
– Used to solve open ended problems of a design or planning
• Backward Chaining
– The inference engine matches the assumed hypothesis or
conclusion which is the goal state with the conclusion or THEN
part
– If the hypothesis is not supported, then the system will attempt
to prove another goal state
– Used for limited in number and well defined problems
– Use classification or diagnosis systems
15. 11- 15
Inferencing Strategies
Conclusion
(Goals)
Input
Data
Few Items Many Possibilities
(For Example, User (For Example, a
Specifications for Computer
a Computer Configuration)
System)
(a) Forward Chaining: IF - Part Matches Shown
16. 11- 16
Inferencing Strategies (Cont.)
Input
Data
Conclusion
(Goals)
Extensive;
Much of the Data Few Possibilities
Obtained by the (Known in Advance
System Querying ((For Example,
the User (For Investment Options)
Example,
Investor’s Profile)
(b) Backward Chaining: THEN - Part Matches Shown
17. 11- 17
How Expert Systems Work:
Uncertainty and Fuzzy Logic
• Resembles human reasoning
• Allows approximate values or inferences and
incomplete or ambiguous data
• Handles uncertainty
• More flexible
• Creative
• Can be used to control manufacturing processes
18. 11- 18
Expert System Technology
• The tool selected for the project must match the
capability of the projected expert system
• Must be able to integrate with other subsystems
and databases
• The tool must match the qualifications of the
project team
19. 11- 19
Expert System Technology
• Specific Expert Systems
– Provide recommendations for a specific task domain
• Expert System Shells
– Shell without a knowledge base
– Furnishes the ES developer with the inference engine, user
interface, and the explanation and knowledge acquisition
facilities
– Domain specific shells
» Incomplete specific expert systems
• Expert System Development Environments
– Run on engineering workstations, minicomputers, or
mainframes
– Integration with databases
• High Level Programming Languages
– LISP, C, C++
20. 11- 20
Expert Systems Technologies
Greater
Complexity of
Greater Higher-Level
Problem and
Flexibility Programming
Environment
Language
Expert System
Development
Environment
Generic Shell
Domain-Specific
Shell
Greater Specific Expert
Ease of Use System
21. 11- 21
Roles in Expert System Development
• Expert
– Knowledge
• Knowledge Engineer
– Knowledge acquisition tactics include interviews, protocol
analysis, observation, and analysis of cases
– Must select a tool with the application of the knowledge
acquisition facility
• User
– End user with a simple shell
– Prototypes are used
22. 11- 22
Development and Maintenance of Expert Systems
1. Problem Identification and Feasibility Analysis
– The problem must be suitable for an expert to solve it.
– Find an expert for the project
– Cost effectiveness must be established
2. System Design and Expert System Technology
Identification
– The system is designed with integration other subsytems and
databases
– Domain knowledge
– Knowledge and inferencing is established with simple cases
3. Development of Prototype
– Knowledge Engineer works with the expert
– Specific Tool is chosen for the project
23. 11- 23
Development and Maintenance of Expert Systems
(Continued)
4. Testing and Refinement of Prototype
– Test with simple cases
– Deficiencies in performance are noted.
– End users test the prototypes.
5. Complete and Field the Expert System
– The interaction with the environment,, users, and other
information systems is tested
– Documented
– User training
6. Maintain the System
– The system is kept current by updating the knowledge base
– Interfaces with other information systems are maintained
24. 11- 24
Development & Maintenance of ESs
Problem Identification and
Feasibility Analysis
System Design and ES
Technology Identification
Development of
Prototype
Testing and Refinement
of Prototype
Yes
Is the Performance
Satisfactory? Complete and
No Field the ES
ES Ready for Use
Maintain ES
25. 11- 25
Expert Systems in Organizations: Benefits
1. An ES can complete its task faster than a human
2. Low error rate, and lower than human error rate
3. ESs make consistent recommendations.
4. ESs are a convenient vehicle for difficult sources
of knowledge
5. ESs bring forth expertise
6. ESs can build organizational knowledge, as
opposed to the knowledge of individuals
7. ESs can be used for training with a faster
learning curve
8. The company can operate an ES in environments
that are hazardous to humans
26. 11- 26
Expert Systems in Organizations: Limitations
1. Limitations of the technology
2. Problems with knowledge acquisition
3. Operational domains as the principal area
4. Maintaining human expertise
28. 11- 28
Overview of Applied Artificial Intelligence
• Natural Language Processing
– Talk to computers and have them “understand”
• Robotics
– Artificial Intelligence, Engineering, and Physiology
– Human like applications
• Computer Vision
– Simulation of the human senses
– Visual scene recognition
• Speech recognition
– Understand speech, also of an unknown speaker
29. 11- 29
Overview of Applied Artificial Intelligence
• Machine Learning
1. Problem Solving Learning
– Accumulate experience about rules
2. Case Based Learning
– collecting cases from a knowledge base
3. Inductive Learning
– Learning from examples
– generate knowledge using rules
30. 11- 30
Applied Fields of AI
AI
Com-
Natural puterized
Expert Computer Machine
Language Robotics Speech
Systems Vision Learning
Processing Recog-
nition
31. 11- 31
Neural Networks
• Computing systems modeled on the human
brain’s interconnected processing elements or
neurons
– 100 billion neuron brain cells
• An array of interconnected processing elements
accept inputs, processing, and then producing an
output imitating the human brain
• Requires sophisticated pattern recognition
• Does not explain the conclusions they make
• Can be made to recognize patterns, and then
apply to new cases
32. 11- 32
Key Terms in Chapter 11
Artificial Intelligence (AI) Expert System Development
Expert System (ES) Environment
Knowledge Base Knowledge Engineer
Heuristic Knowledge Knowledge Acquisition Facility
Knowledge Engineering Natural Language Processing
Knowledge Representation Robot
Rule-Based Expert System Computer Vision
IF-THEN Rule Computerized Speech
Explanation Facility Recognition
Inference Engine Machine Learning
Working Memory Neural Network
Forward Chaining
Backward Chaining
Fuzzy Logic
Expert System Shell
Domain-Specific Shell