Info Vis, Multitasking and Large Displays

1
Info Vis, Multitasking and Large Displays

•  Mary Czerwinski
• Microsoft Research

2
(No Transcript)
3
Information Visualization

• Main ThemeOffload cognitive work to the
  perceptual system
• Focus on animated transitions to maintain context
• Work with George Robertson, Kim Cameron, Daniel
  Robinson

4
Polyarchies

• Visualizing multiple dimensions
  sequentially/simultaneously
• Using animation to orient user spatially
• Questions of timing, freeze frames, linkages,
  cognitive load
• Multiple foci pivots
• Iterative design with internal db stakeholders
  and developers

5
What is the problem?

• Hierarchies are very common
• 20 years of hierarchy visualization RD
• Significant problems remain
• New problems appearing (multiple hierarchies)

6
Current Approaches

• Many 2D and 3D Hierarchy Visualizations
• Each works for some tasks and some scales
• Very few have had user testing
• Windows Tree Control
• Many observed problems

7
Whats wrong with this picture?
8
Problems Cognitive Overhead

• Loss of context
• Or loss of detail
• Separate detail/overview ? extra attentional
  resource required
• Multiple foci is difficult
• Which item is open?

9
Basic View Strategies

• Two view (separate detail/overview views)
• Distorted view
• Distorted data fisheye
• Distorted space 3D, hyperbolic
• Focus in Context (integrated view)
• Many use ANIMATION during transitionsmost not
  empirically evaluated

10
Cone Tree Scales, Integrates Focus Context

• Robertson, Mackinlay Card, Xerox PARC, CHI91
• Limits
• 10 levels
• 1000 nodes
• Up to 10,000
• Animated, complex rotations

11
TreeMap Scales, Integrates Focus Context

• Johnson Shneiderman, U. Maryland, Vis91
• Space filling
• 3000 objects
• MicroLogics DiskMapper

12
Hyperbolic Browser - 1994

• Lamping Rao, Xerox PARC, UIST94
• Projected onto circle, animated
• 1000s of nodes
• Reaction to occlusion problem in Cone Trees

13
Sunburst - 2000

• Stasko Zhang, Georgia Tech, InfoVis 2000
• Radial space-filling
• Techniques for viewing more detail, animated

14
Multiple Hierarchies (3 kinds)

• One hierarchy changing over time
• Time Tube (Chi et al., 1998)
• Taxonomy visualization (Graham et al., 2000)
• MultiTrees (shared subtrees)
• XML3D (Munzner, 1997)
• Polyarchy

15
People and Resources Example

• Multiple Hierarchies Exist
• Direct reporting
• Cost or Profit Center
• Location
• Implicit relationships
• But only one hierarchy shown at a time

16
Polyarchy 1 Selection, 1 Hierarchy
17
Figure . Polyarchy Visualization showing
relationship of three people in the management
hierarchy.
18
Two Styles of Visual Pivot

• Rotating
• Sliding

19
Visual Pivot (Rotation around Vertical Axis)
20
Schematic of Visual Pivot (rotation)
21
Rotation around Horizontal Axis
22
Sliding Animation
23
Stacked (w/links)
24
Animation Controversy

• Tversky et al. (2001) - Animation not always
  useful unless interactive, user controlled
• Robertson, Card Mackinlay (91) rotationsgood
  for maintaining context
• Bartram (98) emergent property of grouping when
  similar motions occur across a dense data display
• Bederson Boltman (98) 1 s. zoom reduced
  errors aided spatial memory

25
Proffitt and Kaiser (93)

• Users analyze animations into relative (rotation)
  and common (translation) motion components
• Designers of animation displays need to recognize
  that moving object configurations interact with
  their displacement perception
• Secondly, rotation and translation motions have
  different perceptual significance
• Rotations define 3D form, while translations
  define observer-relative displacements
• This analysis suggests that the Visual Pivot
  sliding animation may be perceived as
  observer-relative while the rotating animations
  may be perceived as defining 3D form (shows
  relationships but less useful for our tasks?)

26
User Studies

• Study 1 Mockup of visual pivot
• Issues list guided development of prototype
• Study 2 Prototype 2D vs 3D
• Visual Pivot animation was misleading
• Animation speeds were too slow

27
User Studies

• Study 3 Animation Styles and Speeds
• Six animation styles Picked 2 best (sliding and
  rotating)
• Twice as fast as study 2 Still too slow
• Study 4 Prototype 2D vs 3D
• Identified most effective animation style-sliding
• Identified best speed range0.5 sec.
• Study 5 Examined complexity of query and sliding
  v. stacked animations both effective

28
Animation Styles Learning
29
Animation Timing
30
(No Transcript)
31
Unresolved Problems

• Hierarchy ordering
• Server returns siblings in undetermined order
• When 2 or more foci and pivot occurs, 2
  selections may be reversed
• Default is alphabetical ordering
• Text Rendering
• Text rendered in 3D could provide ideal depth
  cues (e.g., change in size depending on distance
  from observer)

32
Notifications Reminders

• Over 2 years of empirical findings-gt principles
  of notification design
• Principles distributed and adopted by many
  notification design teamsongoing
• Ongoing work with Eric Horvitz
• Research to support the design of intelligent
  notifications platforms
• Current work is focused on reminding and
  reinstating context after a task switch

33
(No Transcript)
34
(No Transcript)
35
Messenger Math Problems
36
Attention-Based Principles of Notifications 1

• Unless you are absolutely sure the user wants to
  know what youre telling them at that moment, be
  careful of very salient notifications
• Autoarchive in Outlook
• Frequent audio alerts from messenger
• Users trust is fragile. Once they perceive a
  system is unreliable, it is very hard to win them
  back
• Be cautious repeating information it might be
  outdated or irritating

37
Attention-Based Principles of Notifications 2

• Make notifications situation-awarepresent
  between cognitive chunks
• Early in a task is the worst time to interrupt if
  you want user to remember what they were doing
• When possible, use smart monitoring
• Monitor the user (what are they doing?)
• Content of interruptionrelevant content less
  disruptive, privacy issues
• Demands of current task

38
Women Take a Wider View

• Mary Czerwinski, Desney Tan, George Robertson
• Microsoft Research and CMU

39
Large Display Efforts

• Gender/FOV/Large Display Findings
• Women take a wider field of view to build
  cognitive maps of virtual spaces
• Elegant principle of nav design that benefits
  females without male cost
• Unlocking the code behind principles from
  psychology and good design in navigation tasks

40
Prior WorkNav Gender

• Females known to tend to navigate by landmarks in
  the environment
• Importance of landmarks acknowledged in virtual
  world design (e.g., Darken, Elvins, Vinshon,
  etc.)
• Men known to navigate by broader bearings (e.g.,
  N, S, E W)
• Gender differences often magnified in virtual
  worlds (e.g., Waller, Hunt Knap, 1998)

41
Prior Work--FOV

• Much evidence that restricting FOV leads to
  performance decrements
• Increasing FOV to 90 degrees allows overlapping
  sequence of fixations in memory faster cognitive
  map construction
• Wider FOV results in better eye-hand coordination
  and tracking behavior
• Especially when visual complexity increases
• No gender effects mentioned in literature

42
Experiments 1 2 (CHI 2001)

• Examined novel navigation techniques
• Used large, 36 inch display (Arcturus)
• 2 rear projectors onto a a semi-curved tinted
  Plexiglas surface using Windows multimonitor
  support
• 83 aspect ratio (twice as wide as normal
  displays)
• 36 x 14 inches
• 2048 x 768 pixel display surface
• FOV 75 degrees
• Also smaller, 17 inch display (33 degree FOV)

43
Experiment 1 Test Design

• 17 users (7 female)
• Procedure
• Find, identify, pick up, drop cubes at target
  pads
• Cubes scattered randomly
• Participants placed 4 cubes on 4 pads in each of
  4 conditions, all counterbalanced
• Deadline of 5 minutes
• Testing nav techniques
• Measured trial times

44
World Dimensions

• Tutorial world
• 300 x 300 meters, 4 objects
• Test world
• 500 x 500, 23 objects (tents, roller coasters,
  and rides)
• Both worlds had 4 target cubes and target drop
  pads
• Object was to put 4 cubes on 4 corresponding pads
  as quickly as possible

45
Experiment 2 Conditions

• Chose best nav techniques from Exp. 1
• Exp. 2 3x2 within subjects design

Small Display
Large Display



Basic navigation Speed-coupled flying with
orbit Speed-coupled flying with orbit/glide
46
(No Transcript)
47
Experiments 1 2 Summary

• Larger display may narrow gender gap on
  performance in 3D navigation
• Unanswered questions
• What tasks do they enhance? Why?
• What about them causes better/worse performance?
• Cause of gender effect for navigation tasks?

48
Experiment 3

• Goals--replicate and extend findings from
  Experiment 2
• Hypothesis wider FOV benefits females more than
  males
• Also, better control for display size (all on one
  display)
• DFOV to GFOV ratios identified
• Design
• FOV x display size x gender
• 32.5 v. 75 degree FOVs, 18 36 inches wide
  displays

49
Experiment 3 Methods

• 32 intermediate to advanced PC users (17
  Female)--No 3D gamers
• Avg. age 41 (19 to 60 years old)
• DFOV x GFOV ratios
• Small-narrow 11, small-wide12,
  large-narrow21, and large-wide11
• FOV means GFOV from here on out
• All conditions run on large display

50
Experiment 3 Procedure

• Same task as in Experiments 1 2
• After 4 cubes found, 3 pointing trials
• 1 object and 1 drop pad were removed from world
• Participants had to point at each object from 3
  random locations (spatial memory measure)
• 450 MHz Pentium II Dell computer

51
Experiment 3 Dependent Measures

• Trial time (for all 4 cubes)
• Travel distance
• Travel height (measure of efficiency)
• User satisfaction
• Pointing error
• Kit of Factor Referenced Cognitive Tests MV2 and
  MV3map memory measures

52
Experiment 3 Results

• Map Memory, N.S., t(29)-0.29, p.77
• Performance data
• 2 (gender) x 2 (screen size) x 2 (FOV) repeated
  measures MANOVA
• Percent correct, N.S.
• Main effects
• Gender
• Males faster (193 v. 226 seconds) and flew higher
  (16.5 v. 13.8 meters)

53
Experiment 3 Results

• Main effects continued
• Larger display conditions on avg. resulted in
• Less pointing error (14.8 v. 15.4 meters error)
• Greater distance traveled (6918 v. 5461 meters)
• More flying (15.5 v. 14.9 meters height)
• Faster trial times (205 v. 214 seconds)

54
Experiment 3 Results

• Wider FOVs on avg. resulted in
• Less pointing error (14.8 v. 15.3 meters error)
• Shorter distance traveled (5777.4 v. 6601.7 m.)
• Higher flying (15.8 v. 14.6 meters)
• Faster trial times (199.85 v. 218.7 seconds)
• Planned comparison M-F difference in large
  display, wide FOV condition N.S.

55
Experiment 3 Trial Times
56
Difference between M-F Trial Times
57
Experiment 3 Pointing Error
58
Gender Strategy Differences
59
Experiment 3 Discussion

• User Satisfaction 12/15 males and 14/17 females
  preferred wider fov conditions
• Observed typical overall male superiority
• Large display, wide FOV condition reduces that
  superiority (trial times, pointing error)
• Opposing gender strategies for wide fov conditions

60
Preliminary PrinciplesField of View

• Use a wider field of view (gt75 degrees) coupled
  with a large display (gt36 inches) for better
  female navigability
• Works for both simple and complex information
  spaces (Exp. 4 not reported)
• Ensures females navigate as quickly and
  accurately as males on search and manipulation
  tasks in novel environments
• Could be critical in educational and training
  settings

61
Future DisplaysMass Multiples
62
Future Displays Stanfords I-Room
63
Conclusion

• User research plays pivotal role in developing
  advanced technology _at_MSR
• Leading to better designs
• Identifying new psychological principles
• Blurring the line between basic and applied
  research
• Product teams see value (e.g., big display
  surfaces are hot due to gender findings)