Adaptive User Interfaces

Class One (Sept 23) 
Notes for Paper:
State of the Art in Adaptive User Interfaces
 
Taken by Laconya Ruby
 
 
Approaches to Adaptivity in User Interface Technology:
Survey and Taxonomy
 
T. Kuhme, U. Malinowski, H. Dieterich, M. Schneider-Hufschmidt
 
SIEMENS AG, Munich
Corporate R&D, Systems Ergonomics and Interaction
 
Outline
 
* Introduction
* Design Alternatives
* Models in Adaptive User Interfaces
* Conclusions
 
 
INTRODUCTION:
 
What is an Adaptive User Interface?
"An  _Adaptive System_  is a  _knowledge-based_  system which 
_automatically_  alters aspects of the  _system functionality and 
interface_  in order to accommodate the differing  _preferences and 
requirements_  of  _individual_  system users."  (Benyon 1990)
 
Examples of adaptive behaviour:
     * Selection of interaction techniques and communication 
       channels
     * Task-dependent presentation of forms and menus
     * Task- or user-dependent information presentation
     * Adaptive help
 
 
 
Terms and Relations
 
See Figure 1:  Terms:  Adaptive User Interfaces and Intelligent 
Interfaces (in article)
 
 
 
DESIGN ALTERNATIVES:
 
The Design Space for Adaptive User Interfaces
--> Who takes part in the adaptive process and with which role?
--> Which interaction levels should be adapted?
--> What information should be considered?
--> What is the goal of the envisaged adaptation?
--> When should adaptation occur?
 
 
 
Stages and Agents:
 
Stages:
     Initiative
     Proposal
     Decision
     Execution
 
Agents:
     Designer
     System Administrator
     Local Expert
     User
     System
 
See Figure 2:  Tasks and Agents: Example Configuration (in article)
 
 
 
QUESTION FROM CLASS:
Where is the "evaluation" stage?
REPLY:
The evaluation stage is assumed in the process, it happens after the 
execution and before the following initiative/proposal.
 
 
 
Interesting Combinations
 
See Figure 3:  Classification Scheme:  Tasks and Agents (in article)
 
 
 
Less Sensible Combinations:
 
Lazy Adaptation
 
* *
*
  *
  *
____________________
Dictatorial Adaptation
 
* *
  *
*
*
____________________
Lazy Dictatorial Adaptation
 
* *
* *
*
  *
 
 
 
Example Classifications
 
See Figure 4:  Examples of Classified Systems (in article)
 
 
 
Adapted Constituents
 
See Figure 6:  Adapted Constituents (in article)
 
 
 
Information Considered for Adaptation
 
See Figure 8:  Considered Information (in article)
 
 
 
Goals of Adaptivity
 
See Figure 9:  Goals of Adaptation (in article)
 
 
 
Timing Strategies
 
Changing the interface without the user being in control is 
dangerous
 
Changing the interface before the first session or between sessions is 
preferred to changing during a session - unless it is changed at the 
user's request
 
See Figure 10:  Timing of Adaptation (in article)
 
 
 
Levels and Methods of Adaptation
 
See Figure 13:  Levels of Adaptation (in article)
See Figure 14:  Methods of Adaptation (in article)
 
 
 
MODELS IN AUIs:    
 
Models for Software Structure
* "Embedded" models
	as information source for adaptation
	as object of adaptation
* Software components
* Software architecture
 
 
 
Embedded Models in AUIs
 
See Figure 11:  Elements of the Structural Model of an AUI 
(in article)
 
 
 
User Modeling
 
See Figure 19:  Dimensions of User Models (in article)
See Figure 20:  Knowledge Acquisition (in article)
 
 
 
Task Modeling
 
See Figure 22:  Task Modeling (in article)
 
 
 
Dialog Modeling
 
See Figure 23:  Dialog Modeling  (in article)
 
 
 
Plan Recognition
 
See Figure 24: Plan Recognition (in article)
 
 
 
CONCLUSION:  
 
Conclusions:
--> Taxonomy is useful to identify research gaps and new topics:
	* Amount of necessary information
	* User modeling
	* Timing aspects
	* Design constraints and ergonomic rules on adaptivity
	* Acceptance and evaluation
	* Integration of adaptation components into common user 
	interface tools must be done
	* Overcome restrictions caused by usage of switching as 
	adaptation method
--> Self-Adaptation often fails to match user's needs
--> Computer-Aided Adaptation seems to be a more promising 
    approach
--> A general framework for adaptation is needed
--> Most important: to develop a coherent methodology for the 
    design of adaptive user interfaces
 
 
 
Discussion
 
* How is the term "ergonomic" used in the paper?
Ergonomic is used to mean usability.
 
 
* What are some of the considerations of AUIs?
 
--> Making sure the change helps the user instead of hindering
--> Making sure the changes are consistent
	Consistency is important, but the user can adapt to small 
	changes
 
 
* How do you measure if it is an effective change?
 
--> Time to do a task
Compare the before and after times
 
--> Does it increase the users interest
Emacs example
 
 
* You should have multiple evaluation levels
 
 
* If the user does not have control over the adaptation, he will be 
lost.  The user should know what will be adapted and should not be 
surprised.
 
 
* How do you provide good choices of adaptation to users?
Who decides?
 
-->  Maybe use a critiquing component
computer suggests and the user selects
OR
user suggests and the computer selects
 
This depends on the tasks and the number of users involved
 
--> Maybe it should be a negotiation
but who has the veto power?
 
Again - this depends on the tasks and the number of users involved



-- 
Laconya Dannet Ruby
Georgia Institute of Technology, Atlanta Georgia, 30332
uucp:	  ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!gt6949c
Internet: gt6949c@prism.gatech.edu