29-Oct-24

The Concept of Expertise

• Expertise: extensive knowledge in a narrow
Expert Systems field
and Artificial Intelligence • Expert systems: a computer application that
employs a set of rules based on human
knowledge to solve problems that require
human expertise
• Artificial Intelligence: practical mechanisms
that enable computers to simulate the
reasoning process

1 2

How Do Computers Reason?

The Intelligence of Artificial Intelligence
How do people reason? • Rule-based reasoning: IF-THEN statements
– Categorization represent knowledge encoded as rules
– Specific Rules • Frames: representations of stereotyped
– Heuristics situations that are typical of some category
– Past Experience • Case-based reasoning: adapting previous
– Expectations solutions to a current problem
• Pattern recognition: detecting sounds, shapes
or long sequences

3 4

Other Forms of AI
The Concept and Structure of Expert Systems

• Machine learning – neural networks and • Basic structure of an ES follows the generic
genetic algorithms structure of a DSS
• Automatic programming – mechanisms that • The knowledge base is specific to a particular
generate a program to do a specific task problem domain associated with the ES
(allows non-programmers to “program”) • The main difference between an ES and DSS is
• Artificial life – attempts to recreate biological that the ES contains knowledge acquired from
phenomena within computer-based systems experts in the application domain

5 6

1
 29-Oct-24

Common Expert System Architecture The User Interface in an ES

User
Knowledge • Design of the UI focuses on human concerns
Engineer
such as ease of use, reliability and reduction
User KE
of fatigue
Interface Interface
• Design should allow for a variety of methods
Organization
Systems Inference KE of interaction (input, control and query)
Interface Engine Tool Kit
• Mechanisms include touch screen, keypad,
Knowledge light pens, voice command, hot keys
Base

User Environment Development Environment

7 8

The Knowledge Base The Inference Engine

• Contains the domain-specific knowledge • Here, the knowledge is put to use to produce
acquired from the domain experts solutions
• Can consist of object descriptions, problem- • The engine is capable of performing deduction
solving behaviors, constraints, heuristics and or inference based on rules or facts
uncertainties • Also capable of using inexact or fuzzy
• The success of an ES relies on the reasoning based on probability or pattern
completeness and accuracy of its knowledge matching
base

9 10

The Inference Control Cycle Chaining

Three steps characterize a cycle: • Simple methods used by most inference
1. Match rules with given facts engines to produce a line of reasoning
2. Select the rule that is to be executed • Forward chaining: the engine begins with the
3. Execute the rule by adding the deduced fact to initial content of the workspace and proceeds
the working memory toward a final conclusion
• Backward chaining: the engine starts with a
goal and finds knowledge to support that goal

11 12

2
 29-Oct-24

Forward Chaining Example Backward Chaining Example

Suppose we have three rules: The same three rules:
R1: If A and B then D R1: If A and B then D
R2: If B then C R2: If B then C
R3: If C and D then E R3: If C and D then E

If facts A and B are present, we infer D from R1 If E is known, then R3 implies C and D are true.
and infer C from R2. With D and C inferred, R2 thus implies B is true (from C) and R1
we now infer E from R3. implies A and B are true (from D).
13 14

Building an Expert System

Designing and Building Expert Systems
• Expert System Shells: generic systems that • An early step is to identify the type of tasks
contain reasoning mechanisms but not the (interpretation, prediction, monitoring,
etc.) the system will perform
problem-specific knowledge
• Another important step is choosing the
• Early shells were cumbersome but still allowed experts who will contribute knowledge: It is
the user to avoid having to completely common for one or more of these experts
program the system from scratch to be part of the development team
• Unlike more general information systems
• Modern shells contain two primary modules: a design projects, the software tools and
rule set builder and an inference engine hardware platform are selected very early

15 16

Limitations Associated With ES

Evaluating the Benefits of Expert Systems
Some major benefits: • One important limitation is that expertise is
1. Increased timeliness in decision making difficult to extract and encode.
2. Increased productivity of experts • Another is that human experts adapt naturally
3. Improved consistency in decisions but an ES must be recoded.
4. Improved understanding • Further, human experts better recognize when
5. Improved management of uncertainty a problem is outside the knowledge domain,
6. Formalization of knowledge but an ES may just keep working

17 18

3
 29-Oct-24

Exercise
• You are about to buy a car. Using a rule-based
reasoning technique, describe how you make
your decision

19