CLASS NOTE
==========
Adaptive User Interface - September 26, 1994

Note Taken by: Senis Busayapongchai

Article: Shared Knowledge in Cooperative Problem-Solving Systems
 - Integrating Adaptive and Adaptable Components, Gerhard Fisher

Article Presented by: Susan Farrell


After Susan gave the presentation about the paper, she raised the 
following questions that she thought the class should discussed about:

1. What other problem-solving domain beside design could be enhanced by 
   using malleable system?
2. How could the computer be taught about world knowledge? 
3. What were constraint variables?
4. How could we index huge knowledge?

Students in the class then discussed as follows:

Fault Diagnostics type of application could be enhanced by using 
malleable system.  If something had gone wrong, the system would try to 
adapt and find out the problems.

An Adaptive system should be designed so that it would be workable for 
all types of users - needed to know more than designer or computer had 
all knowledge.  The system should be designed for wide range of users 
from novice to expert.

There was a comment that the brain capacity = 1 year production of mass 
memory.   Computer needed to know several notions e.g. short, tall, 
close, etc. - might use model based AI approach.  However, the computer 
did not have to know everything - just general knowledge - discard 
stupid things.

This article did not explicitly say whether it described function or 
interface issue of adaptive & adaptable systems.

Computer should be able to carry over some general knowledge across 
domains.

Knowledge could be indexed.   For example, In a system used for 
practicing GRE, after you were answering some questions, the system 
could concentrate on particular areas, either your weak or your strong 
areas - could keep asking until you could not answer.

Some touring arrangement systems might not be categorized as adaptive 
system because it only used rules and restrictions - not learning.

When designing or giving criteria for an adaptive system to make some 
changes, you must have some good reasons behind those changes - design 
problem.

Some applications might allow too much customization capability.  It 
looked too much confusing e.g. MS Word 6.0.

The system should be able to determine if the user made mistakes before 
trying to adapt.  Types of users needed to be considered.  Usually 
experts tended to like changing or exploring new things while novice did 
not.

We could first use novice users to test the system.  If these guys 
excepted the system, it had a lot of chances to be exceptable by 
anybody.

==================
Senis Busayapongchai
senis@cc.gatech.edu

-- 
					
					-Senis Busayapongchai (senis@cc.gatech.edu)