1. Artificial Intelligence CAP492
Dr. Souham Meshoul
Information Technology Department
CCIS – King Saud University
Riyadh, Saudi Arabia

2. Artificial Intelligence CAP492
Expert Systems Chapter 8

3. Connectionist Approach
Introduction
Two main approaches for problem solving using knowledge:
Symbolic Approach
Expert systems: Feed the system with knowledge. There is no learning !!
Connectionist Approach
Neural Networks: The system learns from examples by itself.

4. Introduction
Procedural/Object Oriented Programming vs Declarative Programming
PP/OOP: a program instructs the computer -what to do – how to do – in what oreder
Whereas
DP: a program describes what to do without a detailed plan of how to do

5. Symbolic Approach

6. Introduction Expert system
Expert system: knowledge-based systems or rule based-system.
Expert system: a rule based program which encapsulates knowledge from some domain, normally obtained from a human expert in that domain
An expert system simulates a human expert in his/her field of expertise in an attempt to solve a particular problem.
Expert Systems Do Not Replace Experts, But They
Make their Knowledge and Experience More Widely Available
Permit Nonexperts to Work Better
ES acts as a Consultant or Advisor

7. Structure of an Expert System
Major Components
Knowledge base (KB): repository of rules, facts (productions)
working memory: (if forward chaining used)
inference engine: the deduction system used to infer results from user input and KB
user interface: interfaces with user
external control + monitoring: access external databases, control,...
User Interface
Inference
Engine
Knowledge
Base

8. Structure of an Expert System
Knowledge base
The knowledge base contains the knowledge necessary for understanding, formulating, and solving problems
Two Basic Knowledge Base Elements
Facts
Special heuristics, or rules that direct the use of knowledge
Knowledge is the primary raw material of ES
Incorporated knowledge representation
User Interface
Inference
Engine
Knowledge
Base

9. Structure of an Expert System
Inference Engine
The brain of the ES
The control structure (rule interpreter)
Provides methodology for reasoning
User Interface
Inference
Engine
Knowledge
Base

10. Structure of an Expert System
User Interface
Language processor for friendly, problem-oriented communication
Natural Language Processing, or menus and graphics
User Interface
Inference
Engine
Knowledge
Base

11. Structure of an Expert System
Working Memory (Blackboard)
Area of working memory to
Describe the current problem
Record Intermediate results
Records Intermediate Hypotheses and Decisions
1. Plan
2. Agenda
3. Solution
User Interface
Inference
Engine
Knowledge
Base

12. Structure of an Expert System
Explanation Subsystem (Justifier) Area of working memory to
Traces responsibility and explains the ES behavior by interactively answering questions:
-Why?
-How?
-What?
-(Where? When? Who?)
User Interface
Inference
Engine
Knowledge
Base

13. Developing Expert Systems
The Human Element in Expert Systems
Expert
Knowledge Engineer
User
Others

14. Developing Expert Systems
Has the special knowledge, judgment, experience and methods to give advice and solve problems
Provides knowledge about task performance

15. Developing Expert Systems
Knowledge Engineer
Helps the expert(s) structure the problem area by interpreting and integrating human answers to questions, drawing analogies, posing counterexamples, and bringing to light conceptual difficulties
Usually also the System Builder

16. Developing Expert Systems
User
Possible Classes of Users
A non-expert client seeking direct advice (ES acts as a Consultant or Advisor)
A student who wants to learn (Instructor)
An ES builder improving or increasing the knowledge base (Partner)
An expert (Colleague or Assistant)
The Expert and the Knowledge Engineer Should Anticipate Users' Needs and Limitations When Designing ES

17. Architecture of an Expert System

18. Knowledge base: production System
Knowledge is represented using rules of the form :
Rule: if Conditions then Conclusions or
Rule: if Premises then Actions
Rule: if if-part then then-Part
A rule as described above is often referred to as a production rule

19. Knowledge base: production System
Examples:
if symptom1 and symptom2 and symptom3 then disease1
if - the leaves are dry, brittle and discoloured
then - the plant has been attacked by red spider mite
If – it is raining
then – you should take an umbrella
if - the customer closes the account
then - delete the customer from the database

20. Inference Engine: Rule-based reasoning
The essence of a rule-based reasoning system is that it goes through a series of cycles.
In each cycle, it attempts to pick an appropriate rule from its collection of rules, depending on the present circumstances, and to use it.
Because using a rule produces new information, it's possible for each new cycle to take the reasoning process further than the cycle before. This is rather like a human following a chain of ideas in order to come to a conclusion.

21. Inference Engine: Forward Chaining
Forward Chaining is based on Modus Ponens inference rule:
(A, AB ) / B
In other words, if A is true and we have AB then we can deduce that B is true
In the context of Expert System, it is used as follows:
if A is in WM and we have a rule in KB of the form
if A then B then we can deduce B (add B to WM as new information)

22. Inference Engine: Forward Chaining
Do until problem is solved or no antecedents match
Collect the rules whose antecedents are found in WM.
If more than one rule matches
use conflict resolution strategy to eliminate all but one
Do actions indicated in by rule “fired”
Cycles

23. Inference Engine: Forward Chaining
For Conflict Resolution we can use rule-order as an implied priority
Matching
Rules filtering
Conflict Resolution
Execution
Apply the rule
Rules
Rule
Add Then-Part to WM
Cycles

24. Inference Engine: Forward Chaining
Algorithm:
Match WM with KB to select production rules
Eliminate already applied rules
If many rules select one which has the smallest number
Apply selected rule by adding its conclusion to WM
If the problem is solved or no new information added then stop otherwise go to step 1
Matching
Rules filtering
Conflict Resolution
Execution
Apply the rule
Rules
Rule
Add Then-Part to WM

25. Inference Engine: Forward Chaining
Knowledge Base:
Rule 1: If A and B and C then D
Rule 2: if B and C then G
Rule 3: if B and D then F
Rule 4: if B and C and D then E
Rule 5: if A and B then C
Rule 6: if G and D then Z
Inference Engine: Execute many cycles
Working Memory: A, B

26. Inference Engine: Forward Chaining
Inference Engine: Execute many cycles
Cycle 1: Rule 5 → add C
Knowledge Base:
Rule 1: If A and B and C then D
Rule 2: if B and C then G
Rule 3: if B and D then F
Rule 4: if B and C and D then E
Rule 5: if A and B then C
Rule 6: if G and D then Z
Working Memory: A, B, C

27. Inference Engine: Forward Chaining
Inference Engine: Execute many cycles
Cycle 1: Rule 5 → add C
Cycle 2: Rule 1, Rule 2, Rule 5 → add D
Knowledge Base:
Rule 1: If A and B and C then D
Rule 2: if B and C then G
Rule 3: if B and D then F
Rule 4: if B and C and D then E
Rule 5: if A and B then C
Rule 6: if G and D then Z
Working Memory: A, B, C, D

28. Inference Engine: Forward Chaining
Inference Engine: Execute many cycles
Cycle 1: Rule 5 → add C
Cycle 2: Rule 1, Rule 2, Rule 5 → add D
Cycle 3: Rule 1, Rule 2, R 3, R 4, Rule 5 → add G
Knowledge Base:
Rule 1: If A and B and C then D
Rule 2: if B and C then G
Rule 3: if B and D then F
Rule 4: if B and C and D then E
Rule 5: if A and B then C
Rule 6: if G and D then Z
Working Memory: A, B, C, D, G

29. Inference Engine: Forward Chaining
Inference Engine: Execute many cycles
Cycle 1: Rule 5 → add C
Cycle 2: Rule 1, Rule 2, Rule 5 → add D
Cycle 3: Rule 1, Rule 2, R3, R 4, Rule 5 → add G
Cycle 4: Rule 1, Rule 2, Rule 3, Rule 4, Rule 5, Rule 6 → add F
Knowledge Base:
Rule 1: If A and B and C then D
Rule 2: if B and C then G
Rule 3: if B and D then F
Rule 4: if B and C and D then E
Rule 5: if A and B then C
Rule 6: if G and D then Z
Working Memory: A, B, C, D, G, F

30. Inference Engine: Forward Chaining
Inference Engine: Execute many cycles
Cycle 1: Rule 5 → add C
Cycle 2:Rule 1, Rule 2, Rule 5 → add D
Cylce 3: Rule 1, Rule 2, Rule 3, Rule 5 → add G
Cycle 4: Rule 1, Rule 2, Rule 3, Rule 4, Rule 5, Rule 6 → add F
Cylce 5: Rule 1, Rule 2, Rule 3, Rule 4, Rule 5, Rule 6 → add E
Knowledge Base:
Rule 1: If A and B and C then D
Rule 2: if B and C then G
Rule 3: if B and D then F
Rule 4: if B and C and D then E
Rule 5: if A and B then C
Rule 6: if G and D then Z
Working Memory: A, B, C, D, G, F, E

31. Inference Engine: Forward Chaining
Inference Engine: Execute many cycles
Cycle 1: Rule 5 → add C
Cycle 2: Rule 1, Rule 2, Rule 5 → add D
Cylce 3: Rule 1, Rule 2, Rule 3, Rule 5 → add G
Cycle 4: Rule 1, Rule 2, Rule 3, Rule 4, Rule 5, Rule 6 → add F
Cylce 5: Rule 1, Rule 2, Rule 3, Rule 4, Rule 5, Rule 6 → add E
Cylce 6: Rule 1, Rule 2, Rule 3, Rule 4, Rule 5, Rule 6 → add Z
Knowledge Base:
Rule 1: If A and B and C then D
Rule 2: if B and C then G
Rule 3: if B and D then F
Rule 4: if B and C and D then E
Rule 5: if A and B then C
Rule 6: if G and D then Z
Working Memory: A, B, C, D, G, F, E, Z

32. Inference Engine: Forward Chaining
Inference Engine: Execute many cycles
Cycle 1: Rule 5 → add C
Cycle 2:Rule 1, Rule 2, Rule 5 → add D
Cylce 3: Rule 1, Rule 2, Rule 3, Rule 5 → add G
Cycle 4: Rule 1, Rule 2, Rule 3, Rule 4, Rule 5, Rule 6 → add F
Cylce 5: Rule 1, Rule 2, Rule 3, Rule 4, Rule 5, Rule 6 → add E
Cylce 6: Rule 1, Rule 2, Rule 3, Rule 4, Rule 5, Rule 6 → add Z
Cycle 7: Rule 1, Rule 2, Rule 3, Rule 4, Rule 5, Rule 6 → no rule => STOP !!!
Knowledge Base:
Rule 1: If A and B and C then D
Rule 2: if B and C then G
Rule 3: if B and D then F
Rule 4: if B and C and D then E
Rule 5: if A and B then C
Rule 6: if G and D then Z
Working Memory: A, B, C, D, G, F, E, Z

33. Inference Engine: Backward Chaining
Backward Chaining is based on Modus Tollens inference rule:
( B, AB ) /  A
If A implies B , and B is false, then A is false
In other words, if we have AB then proving B is equivalent of proving A
In the context of Expert System, it is used as follows:
To prove a goal B which is not in WM and if we have a rule in KB of the form if A then B then we have just to prove A (A is either present in WM or it exists a rule with A as a conclusion)

34. Inference Engine: Backward Chaining
Consider the following KB and WM:
Inference Engine: Exploration of AND/OR Tree:
Suppose we have the goal: G
To prove G, one can use Rule 2
Proving G is equivalent of proving B and C
As B is present in the WM so it’s true
So remains the proof of C, one can use Rule 5
Proving C is equivalent of proving A and B
As A and B are present in WM (true), we can deduce that C is true
B and C are true then G is true
Knowledge Base:
Rule 1: If A and B and C then D
Rule 2: if B and C then G
Rule 3: if B and D then F
Rule 4: if B and C and D then E
Rule 5: if A and B then C
Rule 6: if G and D then Z
Working Memory: A, B

35. Expert System Shells Shell = Inference Engine + User Interface
ESS allow non programmers to build an expert system, by inserting facts and rules into a generic expert system architecture which is already built.
Some ESSs
BABYLON
JESS ES
CLIPS…..

36. Advantages of expert Systems
An expert system can operate constantly 24 hours per day.
An expert system can exceed the performance of any human expert, as it can combine knowledge from a number of different experts.

37. Limitations of expert systems
Not able to cope with unseen information.
Not able to cope with noise.
Not able to adapt to new environments.
Not able to learn independently in a similar manner that humans learn. They need to be programmed in advance.

38. Conclusions Why use expert systems: Weaknesses:
commercial viability: whereas there may be only a few experts whose time is expensive and rare, you can have many expert systems
expert systems can be used anywhere, anytime
expert systems can explain their line of reasoning
commercially beneficial: the first commercial product of AI
Weaknesses:
expert systems are as sound as their KB; errors in rules mean errors in diagnoses
automatic error correction, learning is difficult
the extraction of knowledge from an expert, and encoding it into machine-inferable form is the most difficult part of expert system implementation