WHAT IS ARTIFICIAL INTELLIGENCE?

Download Report

Transcript WHAT IS ARTIFICIAL INTELLIGENCE?

EXPERT SYSTEMS
Part I
What is there in an expert
system?






Components of an expert system
Knowledge acquisition and knowledge
engineering
Explanations
User interface
Expert system shells
Expert systems demos
2
Introduction
Expert systems solve real problems which
normally would require a specialized human
expert (such as a doctor or a lawyer).
Knowledge characteristics:
• heuristic in nature,
• based on ``rules of thumb''
3
Domains
• Medicine
• Engineering
• Geology
• Computer Science
• Business
• Law
• etc
4
Types of problems



diagnosis (e.g., of a system fault,
disease or student error)
design (of a computer systems, hotel
etc)
interpretation (of, for example,
geological data
5
Rule Based Production
Systems



Most ES are rule-based production systems
Their knowledge is represented in the form
of production rules to provide diagnoses or
advice on the basis of input data.
Production rules are if-then rules
 IF
it rains, then you will get wet
 IF
it rains, then you should wear a coat
6
Components of an expert
system
User
Expert system shell
User
Inter
face
Explanation
system
Case specific
data: Working
storage
Inference
engine
Knowledge base
Knowledge
base editor
7
Expert system knowledge base

Domain inference rules and object
descriptions

Case specific data

Knowledge base editor: creating and
updating the knowledge base
8
Knowledge acquisition and
knowledge engineering

Domain expert

Knowledge engineer

Knowledge acquisition : The process
of "extracting" the knowledge from the
domain expert and representing the
knowledge in a suitable form that can be
used by the expert system
9
10
Inference Engine

Forward chaining: derive all conclusions given
certain facts






Select applicable rules
Choose a rule (conflict resolution)
Execute the rule and update the casespecific data
Backward chaining: prove a given goal by
verifying the conditions in the IF-THEN rules
Coping with uncertainty
Mixed reasoning
11
Forward Chaining


Forward chaining is a reasoning model
that works from a set of facts and rules
towards a set of conclusions, diagnoses or
recommendations.
When a fact matches the antecedent of a
rule, the rule fires, and the conclusion of
the rule is added to the database of facts.
12
Conflict Resolution



More than one rule will may be used
A conflict resolution strategy decides which rule
to use.
Strategies:
give rules priorities and use the conclusion
that has the highest priority.
 apply the rule with the longest antecedent
 apply the rule whose condition was most
recently added to the database.

13
Rete Algorithm for Forward
Chaining



Reorganize the rules in a network so that
the conditions are linked to the rules,
i.e. rules are indexed by their
antecedents
When a new conclusion is made or a new
fact arrives, examine only those rules
whose antecedents contain the new
conclusions or facts
The algorithm makes forward chaining
much faster at the expense of greater
memory requirements
14
Backward Chaining



Used when a particular conclusion is to
be proved.
Works back from a conclusion towards the
original facts.
When a conclusion matches the conclusion
of a rule in the database, the system tries
to prove the antecedents of the rule.
15
Coping with uncertainty


The ability of the system to reason
with rules and data which are not
precisely known.
Certainty factors:
 Apply to the known facts: how
much we trust the recorded facts
 Apply to the rules: how much we
trust the rules
16
Meta Rules

Meta rules : knowledge about how the
system works.

Example of meta rules:


The rules that determine the conflict
resolution strategy
Meta rules might define that knowledge
from Expert A is to be trusted more than
knowledge from Expert B.
17
Meta Rules

Implementation of meta rules:
 Implicit – in the inference engine
(e.g. conflict resolution)
 Explicit
- treated by the system like
normal rules, but are given higher
priority
18
Mixed Reasoning


Collect initial data and use forward
reasoning to determine a set of
possible conclusions
Use backward reasoning to prove or
disprove each conclusion
19
Example of Mixed Reasoning in
Medical ES






Collect data to build hypotheses
For each hypothesis H:
If H is in the facts database, it is proved.
Otherwise, if H can be determined by asking a
question, then enter the user’s answer in the
facts database. Hence, it can be determined
whether H is true or false.
Otherwise, find a rule whose conclusion is H. Now
apply this algorithm to try to prove this rule’s
antecedents.
If none of the above applies, we have failed to
prove H.
20
Explanations

Why a particular question is asked

How a conclusion has been reached

Why not some other conclusion

What if a particular symptom is present

Terminology – explain the terms used
by the system
21
User interface



End-users
Knowledge engineer
Experts
Requirements:
 Self-consistent
 Consistent with widely used interface systems
 Provide navigation help
 Incorporate user model
22
Expert System Shells

The shell is a piece of software which contains

the user interface,
a format for declarative knowledge in the
knowledge base, and
 an inference engine.


The knowledge engineer uses the shell to
build a system for a particular problem
domain.
23
CLIPS
CLIPS is C Language
Integrated Production
System –
an expert system shell.

CLIPS uses a
LISP-like notation
to enter rules.

24
CYC



A frame based production system.
Uses a database of over 1,000,000 facts
and rules, encompassing all fields of
human knowledge.
CYC can answer questions about all kinds
of knowledge in its database, and can
even understand analogies, and other
complex relations.
25