2. EXPERT SYSTEMS AND AI
• Currently there is no such thing as Artificial
Intelligence – as you see in movies
• Computers cannot think like humans
• Expert systems are special database systems
that mimic the knowledge and reasoning of
humans
• They work in a very clearly defined area
based on the rules and data entered
3. EXPERT SYSTEMS HAVE THREE/FOUR PARTS
User
interface
Inference
engine
Knowledge
Base
Rules Base
4. USER INTERFACE
• The user interface allows the system to be
queried/questioned by the user
• The system “asks questions” of the user in a
variety of ways
• Typed questions
• Automatic data collection via sensors
• Scanned material
• The system provides recommendations to the
user based on the rules and data in the system
• The user must then make a decision to use
this advice or not
5. USER INTERFACE
• The user interface allows the expert system to
communicate with the user
6. INFERENCE ENGINE
• The Inference engine is the part that ties
the system together
• The Inference engine analyses the inputs
from the user interface based on a set of
pre-programmed rules (rules base) and
using the data (knowledge base)
• Based on all of this the system makes a
recommendation
7. THE KNOWLEDGE BASE
• To create a knowledge base Information
from human experts is collected
• This information is entered into the system,
organised and then stored
• This information is what the system uses to
answer questions and make decisions
• Therefore it is vital that the data is accurate
and up-to-date
8. THE RULES BASE
• The rules base is constructed using
approved rules that apply to known facts to
deduce new facts
• The rules base use many if-then statements
9. AN EXAMPLE OF USING “RULES”
• A dichotomous key will give you an idea about
how the inference rules can be used to arrive at
an answer.
• Dichotomous keys are
often used in Biology of
identification
10.
11.
12. USES
• Diagnosis of medical illnesses
• Searching for oil and mineral reserves
• Chess playing robots
• Gaming
• Car engine fault diagnosis
13. ADVANTAGES
• Can complete tasks and sort through possibilities
much faster than humans
• Error rates can be very low – lower than for a
human
• Recommendations are always consistent given the
same data
• Recommendations are impartial – have no human
judgments or emotions
14. DISADVANTAGES
• Are very expensive to create
• They do not learn from their mistakes – have no
common sense
• Difficult and time consuming to construct –
including gathering of data and construction of
rules
• The knowledge base needs to be updated when
new knowledge becomes available
• Training of staff to use and maintain this new IT
system
15. EXAM QUESTION: WHAT ARE THE STEPS IN CREATING
AN EXPERT SYSTEM?
1. Interview experts for data
2. Design the Knowledge Base
3. Design the Rule Base
4. Design the Interface – I/O
5. Choose software and construct the Expert
System
6. Test the Expert System
7. Check with experts that the system is giving
sensible answers
8. Document the system
16. EXAM QUESTION: WHAT ARE THE STEPS IN USING AN
EXPERT SYSTEM (EXAMPLE AN ANIMAL VET)
1. The expert system would “ask” all the necessary questions
2. The VET would input the answers (symptoms)
3. The inference engine would consult the rule base as to what to
do
4. The inference engine would be used to search the knowledge
base for appropriate data
5. The symptoms would be analysed
6. Suggested treatment/advice would be output to the VET
7. The VET would decide based on the information what treatment
to use