What%20are%20your%20interactions%20doing%20for%20your%20visualization?

1
What are your interactions doing for your
visualization?

• Remco Chang
• UNC Charlotte
• Charlotte Visualization Center

2
Outline Three Areas of Proposed Research

• What is the role of interaction in visual
  analytics?
• Is there a science to designing interactions and
  applying them to visualizations?
• How do we know if an interaction is good?
• Can we evaluate (quantify?) the benefits (or
  costs) of interactions?
• If analysts use interactions to perform analysis,
  can we store the knowledge in the interactions?
• Is it possible to create a knowledge-base by
  extracting knowledge from interaction logs?

3
IntroductionRole of Interaction

• Most people in the visual analytics community
  believe that interactivity is essential for
  analysis
• A visual analysis session is more of a dialog
  between the analyst and the data the
  manifestation of this dialog is the analysts
  interactions with the data representation
  Thomas Cook 2005
• Without interaction, a visualization technique
  or system becomes a static image or autonomously
  animated images Yi et al. 2007

4
MotivationRole of Interaction

• More explicitly Pike et al. 2009
• A central precept of visual analytics is that it
  is through the interactive manipulation of a
  visual interface the analytic discourse that
  knowledge is constructed, tested, refined, and
  shared.
• These visual displays must be embedded in an
  interactive framework that scaffolds the human
  knowledge construction process with the right
  tools and methods to support the accumulation of
  evidence and observations into theories and
  beliefs.

5
VISUAL analytics or visual ANALYTICS?

• Observation Current designs of visual analytical
  systems start with visual representation and add
  in appropriate interactions afterwards.
• Visual analytics visual representation
  analytics
• assume that (interaction analytic discourse),
  then
• Visual analytics visual representation
  interaction
• VISUAL analytics VISUAL REPRESENTATION
  interaction
• visual ANALYTICS visual representation
  INTERACTION
• Proposal If we start the design of visual
  analytical systems with interactions (i.e., how a
  user would perform a series of tasks, or to
  generate hypotheses), we could focus on the
  ANALYTICS aspect in the design.
• This seems pretty hard to do Arguably because
  we dont really understand the nature of
  interactions for the purpose of analytics.

6
Case Study Brushing and Linking

• The linchpin in most visualizations that utilize
  multiple coordinated views.
• Spotfire, GeoVISTA, JIGSAW, etc.
• However, when used in a collaborative
  environment, its purpose becomes slightly
  different even though the implementation is
  (mostly) the same. Isenberg et al. 2009
• Hypothesis the nature of Brushing and Linking is
  to coordinate between different perspectives of
  the same data elements, especially for data of
  high dimensionality.
• It is now easier to consider a system design
  around this

Visual
Analytics
7
EvaluationThe Benefits of Interactions

• Scientifically, how is interaction useful?
• With interaction,
• Does an analyst perform tasks faster?
• Does an analyst perform tasks more accurately?
• Short answer no
• Lim et al. 1996 Jeong et al. 2009a Jeong et
  al. 2009b Lipford et al. 2009

8
Evidence Interaction Is Useful in
Visualizations

• Empirical evidence that interactivity is useful
• (1) Users dont give up as easily Jeong et al.
  2009

Green bar denotes the number of participants who
gave up during an analysis. (iPCA is an
interactive visualization, and SAS/INSIGHT is a
traditional text-based interface with limited
interactivity)
9
EvidenceInteractivity Is Useful in
Visualizations

• Empirical evidence that interactivity is useful
• (2) Users become more proficient faster
• The longer a user uses an interactive
  visualization, the better (faster) they become.
• Whereas when the same user uses a non-interactive
  visualization, the amount of time spent remains
  (roughly) the same.

Slow
Fast
10
EvidenceInteractivity Is Useful in
Visualizations

• Empirical evidence that interactivity is useful
• (3) Users prefer interactivity Jeong et al.
  2009

Users giving letter grades to the two tools after
using them during an experiment. (iPCA is an
interactive visualization, and SAS/INSIGHT is a
traditional text-based interface with limited
interactivity)
11
Future WorkHow is Interactivity Useful?

• We propose that
• Interactivity is indeed useful
• Weve been measuring the wrong things
• Hypothesis
• Interactivity is useful to keep a user in a
  cognitive zone
• which is why they dont give up
• Interactivity allows the user to gather more
  contextual information
• users spend more time to understand the problem
  before attempting to solve it
• We need new metrics and methods to measure the
  benefits of interactivity

12
ProvenanceCapturing User Interactions

• What is in a users interactions?
• If (interactions analytic discourse), what can
  we learn from the users interactions?
• Is it possible to extract analysis from
  interactions?

13
StudyWhat is in a Users Interactions?

• Goal determine if there really is analysis in
  a users interactions.

Grad Students (Coders)
Compare! (manually)
Analysts
Strategies Methods Findings
Guesses of Analysts thinking
Logged (semantic) Interactions
WireVis
Interaction-Log Vis
14
ResultsWhats in a Users Interactions

• From this experiment, we find that interactions
  contains at least
• 60 of the (high level) strategies
• 60 of the (mid level) methods
• 79 of the (low level) findings

R. Chang et al., Recovering Reasoning Process
From User Interactions. IEEE Computer Graphics
and Applications, 2009. R. Chang et al.,
Evaluating the Relationship Between User
Interaction and Financial Visual Analysis. IEEE
Symposium on VAST, 2009.
15
ProvenanceFuture Work

• Using semantic interaction capturing, we might be
  able to collect all the analysis processes of
  expert analysts and create a knowledge-base that
  is useful for
• Training many domain specific analytics tasks
  are difficult to teach
• Guidance use existing knowledge to guide future
  analyses
• Verification and validation check for accuracy
  and correctness
• But our study was crude and made lots of
  assumptions
• How do we extract analysis from interaction logs
  semi-automatically?
• Can these methods be generalized to all
  visualizations?
• What does a knowledge-base of interactions look
  like and how to use it?
• A model of how and what to capture in a
  visualization for extracting an analytical
  process is necessary.

16
Conclusion

• We do not yet have a perfect foundation on the
  science of interaction, but we are getting
  there.
• The three areas that I propose that would have
  the highest impact in interaction research are
• Fundamental (Functional) understanding of
  interaction and interaction techniques
• Evaluation methods and metrics for measuring the
  benefits (and costs) of interactions
• Capturing and re-using interactions to create a
  knowledge-base of analysts strategies and methods

17

• Thank you!
• rchang_at_uncc.edu
• http//www.viscenter.uncc.edu/rchang

18
Backup Slides
19
ResultsWhats in a Users Interactions

• Why are these so much lower than others?
• (recovering methods at about 15)
• Only capturing a users interaction in this case
  is insufficient.

20
Understanding Interaction as a Science

• With deeper understanding of

21
Scenario Revisited

• Task design a visual analytical system to
  analyze IP logs, starting the design with
  interaction elements.
• Solution A system that displays different
  aspects of the IP data (e.g., dest IP, orig IP,
  time, port number, etc.) that are coordinated
  through Brushing and Linking.
• Visual displays (integrated or coordinated
  multi-views) is a secondary consideration.
• The visual representation of what IP logs could
  be can also be considered independently.
• (Number of users and the applied environment)?

22

• Taxonomy of Interaction Techniques Yi et al.
  2007
• Select mark something as interesting
• Explore show me something else
• Reconfigure show me a different arrangement
• Encode show me a different representation
• Abstract/Elaborate show me more or less detail
• Filter show me something conditionally
• Connect show me related items
• Seems all 7 elements are necessary, but that
  itself doesnt lead to a design of a system

23
Is Visual ANALYTICS possible?

• We propose that indeed it is