[PPT] - The Need for an Interaction Cost Model in Adaptive Interfaces Bowen PowerPoint Presentation

SLIDE 1

The Need for an Interaction Cost Model in Adaptive Interfaces

Bowen Hui, Sean Gustafson, Pourang Irani, Craig Boutilier

Department of Computer Science University of Toronto and University of Manitoba

Advances in Visual Interfaces (AVI’08)

1

SLIDE 2

Need for Software Customization

Increasing complexity

– Lost in interface/functionality – Repeated customization effort

Most affected users

– People with cognitive, sensory, motor impairments – Elderly people – Children – Novices

2

SLIDE 3

Intelligent Interfaces

Design objectives

– Minimize user effort – Maximize ease of interaction

Existing implementations:

– Auto-completion – Toolbar suggestions – Adaptive menus (add/hide/move) – Etc.

3

SLIDE 4

Research Objectives

Account for existing interaction factors
Predict costs/benefits of interaction
Explain individual differences

4

SLIDE 5

Decision-Theoretic Framework

Actions lead to outcomes probabilistically
Impact of intelligent actions
Tradeoffs between costs and benefits
Maximizing (long-term) expected utility

5

SLIDE 6

Impact of actions:

Utility of Customization Actions

Action Savings Processing Occlusion Bloat Disruption Interruption

AUTO

X X

TOOLBAR

X X X X

ADD

X X X X

HIDE

X X X X

MOVE

X X X

HINT

X X X X

ASK

X X X X

6

SLIDE 7

Utility of Customization Actions

Compute utility of each interaction factor
Overall Utility = w1utilityfactor1 + w2utilityfactor2 + …
Each component models:
Objective value
Subjective utility

7

SLIDE 8

Utility of Customization Actions

Compute utility of each interaction factor
Overall Utility = w1utilityfactor1 + w2utilityfactor2 + …
Each component models:
Objective value
Subjective utility

Models existing interaction factors Predicts costs/benefits of interaction Models individual differences

8

SLIDE 9

Interaction Cost Model

Predictive model of interaction factors
Savings
Information processing
Occlusion
Bloat
Disruption
Interruption

9

SLIDE 10

Quality = GOMS(Steps, Mode)

Model of Savings

Steps Mode Quality Frustration Benefits

f

Savings Independence Distractibility Neediness

10

SLIDE 11

Expertise
ProcessTime = Hick-Hymann(Length)

if expert

= Visual_Search(Length)

if naive

Model of Processing

Length Process Time Expertise Cost of Processing

11

SLIDE 12

Model of Occlusion

. . .

?

12

? ? Overlap Frustration Cost of Occlusion Distractibility

SLIDE 13

Model of Bloat

? ? Excess Feature Tolerance Cost of Bloat Distractibility

?

. . .

13

SLIDE 14

Occlusion Experiment

Direction, Size, Opacity, Proximity, Intersection
Task completion time
12 participants

14

SLIDE 15

Analysis Techniques

Factor analysis

– Identifies most relevant variables

ANOVA

– Finds significance among means of different users

F-test

– Determines minimal model complexity required

15

SLIDE 16

Model of Occlusion

Opacity Blocked Overlap Frustration Cost of Occlusion

16

SLIDE 17

Objective Occlusion Function

Overlap = f(Blocked, Opacity)
Blocked=0:
overlap = constant
Blocked=1:
Cubic in Opacity, for half of the users
Linear in Opacity, for remaining users

Opacity Blocked Overlap

17

SLIDE 18

Bloat Experiment

Shown, Used
Task completion time
12 participants

18

SLIDE 19

Model of Bloat

Used Shown Excess Feature Tolerance Cost of Bloat Distractibility

19

SLIDE 20

Objective Bloat Function

Unused = Shown - Used
Excess = f(Unused)
Linear, for most users
Quadratic, for 1 user
Cubic, for 1 user

Used Shown Excess

20

SLIDE 21

Simulations

Markov decision process (MDP)
Adaptive menu
Actions: add/delete menu item or do nothing
Utility = w1Bloat + w2Savings
Bloat = f(Excess, Feature Tolerance, Distractibility)
Savings = f(Quality, Frustration, Neediness,

Distractibility, Independence)

21

SLIDE 22

MDP for Adaptive Menu

Used Shown Quality Feature Tolerance Bloat Distractibility Frustration Independence Neediness Savings

22

SLIDE 23

MDP for Adaptive Menu

Feature Tolerance Used Shown Quality Bloat Distractibility Frustration Independence Neediness Savings

A

23

SLIDE 24

MDP for Adaptive Menu

Feature Tolerance Used Shown Quality Bloat Distractibility Frustration Independence Neediness Savings

A

24

SLIDE 25

MDP for Adaptive Menu

Feature Tolerance Used Shown Quality Bloat Distractibility Frustration Independence Neediness Savings

A t t+1

25

SLIDE 26

Results: Effect of Bloat

Distractibility Tolerance Shown Policy Low/medium Feature-keen Any Add High Feature-keen Few Add Low Feature-shy Many Delete

ther
ther
ther

No action

26

SLIDE 27

Results: Individual Adaptation

Distractibility Tolerance Shown Policy Low Keen/shy Any Add Medium/high Feature-keen Any Add Distractibility Tolerance Shown Policy Low Feature-keen Any Add Low Feature-shy Many Delete Medium Feature-shy Many Delete

Most receptive user:
Least receptive user:
Do nothing for all other cases

27

SLIDE 28

Summary and Future work

Decision-theoretic framework for adaptive

interfaces

Formal model for interaction costs
Systematic analysis
Models individual differences
Simulation as proof of concept
Usability evaluation (next)

28