Transcript The 3rd Dimension

SIMS 247: Information Visualization
and Presentation
Marti Hearst
Oct 10, 2005
1
Today
• Another animation example
– Fluid documents
• The third dimension
– The information visualizer and follow-ons
– Navigating in 3D space
– The debate: does 3D help?
• Cognitive abilities and 3D
• 3D vs 2D
– Alternative:
• 2.5D sequences (Automatic generation of assembly
instructions)
2
Fluid Documents
•
Zellweger, Chang, Mackinlay. Fluid links for informed and incremental
hypertext browsing. CHI'99 Extended Abstracts
• Better document-reading interface
• Combines
– Lightweight animation
– Focus + context
3
3D + Animation
• Pioneered by Card and Robertson
– Had state-of-the-art graphics hardware; wanted to
see what happens when pushing the boundaries
– Motivated by Card & Moran’s theories of cognitive
architecture
•
•
•
•
Information Visualizer (PARC)
WebBook/Webforager (PARC, 1996)
Data Mountain (MS Research, 1998)
Task Gallery (MS Research, 2000)
4
Information Workspaces
• Objective:
– Decrease the costs for performing informationintensive tasks, or, alternatively, increase the scope
of information that can be utilized for the same cost.
• Method:
– Large Workspaces
– Make the immediate workspace virtually larger
– Real-Time Interaction
– Maximize the interaction rates
– Visual Abstractions
– Speed assimilation and pattern detection
8
5
Information Visualizer
17
6
research.microsoft.com/~ggr/gi97.ppt
Web Forager
http://research.microsoft.com/ui/TaskGallery/index.htm
24
7
Data Mountain
• Robertson, Czerwinski et al, 1998
• Follow-on to Information Visualizer
• Organizing bookmarks
using pile metaphor
• Uses:
–
–
–
–
Spatial organization
3D view with 2D interaction
Cartoon animation details
Subtle audio cues
• Debate:
– Is this better than 2D?
8
Task Gallery
Robertson et al., 2000
http://research.microsoft.com/ui/TaskGallery/index.htm
23
9
3D vs. 2D
• Cockburn & McKenzie ’02
– Results for prior work with 3D systems are
primarily negative for viz of things that are not
inherently in 3D, but really results are mixed
– Compared 2D, 2½D and 3D views of web page
thumbnails
– Did this for both physical and virtual interfaces
– Compared sparse, medium, and dense displays
10
3D vs. 2D: Cockburn & McKenzie ’02
11
3D vs. 2D
• Cockburn & McKenzie ’02
– Results:
• Time taken sig. increased through 2D -> 3D
interfaces
• Subjective assessment sig. decreased 2D -> 3D
• Performance degraded with denser problems
• 3D virtual interface produced the slowest times
• People prefered the physical interfaces
• People were better at using their spatial memory
than they expected to be
• There was a problem with the physical 2½D
display
12
Interacting with 3D spaces
• Path-drawing for 3D walkthrough, Igarashi et
al, UIST ’98
• Problem: interacting with 3D via 2D screens
• Solution: be clever about how to convert 2D to
3D based on what the user is likely to intend
13
The Role of Cognitive Abilities
• Leitheiser & Munro ‘95
– Summarizes the results of earlier psychological
research on spatial aptitiude
– Also summarizes work on effects of spatial aptitude
and UI use
– Presents a study comparing a GUI with a command
line interface, taking spatial abilities into account
14
The Role of Cognitive Abilities
• Leitheiser & Munro ’95
• Hypotheses:
– Users with high spatial ability would benefit more from the
GUI than those with low spatial ability (H1)
– Users with high verbal ability would perform better on
command line interfaces (H2)
• Tasks:
– Obtain system time, list files, look up a file update time,
open a subdirectory, move a file, copy a file, etc
– Between subjects GUI (Mac) vs. Command line (DOS)
• Findings:
– H1 supported
– H2 not supported
– Everyone did better on the GUI
• Low spatial ability users using the GUI required 90% of the
time needed for command line interface
15
Infovis Meta-Analysis
(Empirical studies of information visualization:
a meta-analysis, Chen & Yu IJHCS 53(5),2000)
• Goal
– Find invariant underlying relations suggested
collectively by empirical findings from many
different studies
• Procedure
– Examine the literature of empirical infoviz studies
• 35 studies between 1991 and 2000
• 27 focused on information retrieval tasks
• But due to wide differences in the conduct of the
studies and the reporting of statistics, could use only
6 studies
16
Infovis Meta-Analysis
(Empirical studies of information visualization:
a meta-analysis, Chen & Yu IJHCS 53(5),2000)
• Conclusions:
– IR Infoviz studies not reported in a standard format
– Individual cognitive differences had the largest effect
• Especially on accuracy
• Somewhat on efficiency
– Holding cognitive abilities constant, users did better
with simpler visual-spatial interfaces
– The combined effect of visualization is not
statistically significant
17
Gender differences and 3D
• Previous studies often found gender
differences in 3D navigation
• Czerwinski et al. wondered why; saw a hint in
one study, did a followup study in detail
• Idea: change the assumptions
– Make screen wider
– Gender performance differences disappear
– Both improved
18
Sequences of Steps vs
3D + Animation
• Heiser, Phan, Agarwala, Tversky, Hanrahan ‘04
• Domain: assembly instructions
• Identify
– How people conceive of 3D assemblies
– How people comprehend visual instructions
• Validate
– Build automated instruction design system
– Evaluate usability of resulting instructions
Slides from Heiser et al.
19
Ensure Visibility of Parts
•
•
•
•
Show parts added in each step
Show mode and location of attachment
Avoid changing viewpoint
Use physically stable orientation
Slides from Heiser et al.
20
Illustrate Assembly Operations
Structural diagrams
Action diagrams
• Use action diagrams rather than structural
• Use arrows and guidelines to indicate attachment
Slides from Heiser et al.
21
Usability Study
• 30 Participants
• Given 1 of 3 instruction sets: hand-drawn, factory, computer
• Assemble TV stand using instructions
Slides from Heiser et al.
22
Instructions Tested
Hand-drawn
Computer generated
Slides from Heiser et al.
Factory
23
Results
25
Time to
assemble
(min)
20
15
18.9
16.0
10
10.2
5
0
Hand-drawn
•
•
Factory
Computer
Errors:
Hand-drawn 1.6 Factory 0.6 Computer 0.5
Users rated task as easiest in computer condition
Slides from Heiser et al.
24
Next Time
• Martin Wattenberg!
– SIMS distinguished lecture: 4-5:30pm
– Our class too
• Some of Martin’s Projects:
–
–
–
–
–
–
Map of the Market
Baby names visualizer
Music arcs
History flow
Bioinformatics visualization
Conversation visualization
25