Transcript Yimeng Dou

Dynamic Visualization
Dynamic Queries For Visual Information
Seeking by Ben Shneiderman
Data Visualization Sliders by Stephen G.
Eick
Presented by Yimeng Dou
05-21-2002
[email protected]
Overview
Dynamic Queries
Applies the principles of direct
manipulation to the database
environment
Data Visualization Sliders
Dynamic Query Approach
… rapid, animated, visual display of search results
 Visual presentation of the query’s components
 Visual presentation of results
 Rapid, incremental and reversible control
 Immediate and continuous feedback
 No need to learn a specific query language to
use dynamic query.
Example 1: HomeFinder System
HomeFinder
 Christopher Williamson's HomeFinder showed
a map of Washington, DC and 1100 points of
light indicating homes for sale.
 Users could mark the workplace for both
members of a couple and then adjust sliders to
select circular areas of varying radii.
 Other sliders selected number of bedrooms
and cost, with buttons for air conditioning,
garage, etc. Controlled experiments with
benchmark tasks showed dramatic speed-ups
in performance and high subjective
satisfaction.
An Improvement To HomeFinder
ProgramFinder
plots the available
programs on a
map of Maryland.
Adjusting the
controls updates
the display which
shows a dot for
each program that
matches. A click
on a program
provides more
details and the
press of a button
generates the
paperwork.
Example 2: Cancer Rate
 (Cervix cancer rates are color coded on the
map.The year slider shows time trends. )
 The other sliders allow interactive filtering of
the gap according to the three chosen
demographic parameters.
Example 3: Periodic Table
 Periodic table with chemical symbols in red
and six sliders for attributes such as atomic
radius, ionization energy, and electronegativity.
 As users move the sliders, the chemical
symbols change to red showing the clusters,
jumps, and gaps that chemists find fascinating.
 A study with 18 chemistry students showed
faster performance with use of a visual display
(versus a simple textual list) and a visual input
device (versus a form fillin box).
Example 3: Periodic Table
Example 4:Tabular Display
 When there are no natural graphical displays
for the output, dynamic queries can still be
implemented with result sets shown in a
traditional alphanumeric tabular display.
 The sliders and buttons are created semiautomatically by the program depending on
the values that exist in the imported ASCII
database.
 Display is updated only when the user releases
the mouse button.
Tabular Display
Example 5: Dynamic Query of Unix
Directory
 Sliders for size (in kilobytes) and age (in days)
of files enabled 18 users to answer ten
questions such as "How many files are younger
than umcp_tai?"
 The three versions of the program are:
highlighting matches with color
highlighting matches with asterisks
displaying only the matching lines
 In five of the tasks there was a statistically
significant speed advantage for the
Expand/contract interface.
Example 5: Dynamic Query of Unix Directory
Advantages
 Visual presentation of query components
 Visual presentation of results
 Rapid, incremental and reversible actions
 Selection by pointing (not typing)
 Immediate and continuous feedback
 For data in which there is a known relationship
among variables, the dynamic queries interface
is useful for training and education by
exploration.
 Where there is so much data, dynamic queries
may help users to discover patterns, form and
test hypotheses
Disadvantages
…stem largely from their poor match with current
hardware and software systems
 Requirement for rapid performance in search
algorithms and display strategies cannot be easily
satisfied with current database management tools.
 Application specific programming is needed to take
the best advantage of dynamic query methods.
 Current dynamic queries implement only simple
queries that are conjunctions of disjunctions, plus
range queries on numeric values.
 Visually handicapped and blind users will have a more
difficult time with these widgets and outputs
Research Directions (1)--Database and display
algorithms
For small database (main memory): array
indexing, grid structures, quad trees and k-d
trees.
Larger database (disk): R-trees, grid files,
various B-trees.
Display algorithms which can update quickly.
(Buckets adjusted to granularity of the
slider)
Data compression methods
Screen management algorithms
Only repaint the area that is changed
Manipulation of the palette by color
indexing can be effective for irregularly
shaped regions
Research Directions (2)--User Interface Design
Use pictures and capitalize on the human
visual system.
Appropriate coding of properties like size,
position, shape and color, to reduce explicit
selection.
Graphical display properties such as color
Auditory properties.
No natural two-dimensional representation
of the data is in currently available widgets.
Needed: New Widgets
 Existing widgets are poorly matched with the
needs of expert users. One possible solution:
two dimensional input widgets. (Only one
selecting is required to set two values, correct
selections can be guaranteed).
 Three and higher dimensional input widget
may facilitate the exploration of complex
relationships. One possibility: 3-D Mouse.
 Ways of specifying alphanumeric fields. One
possible solution: alphaslider.
 How to specify complex boolean combinations
of attributes. Example: filter/flow model.
Filter/Flow model
 Users can select from the
set of attributes and get
an appropriate filter
widget
 (type-in for interest areas,
sliders for cost, and buttons
for scholarships)
 The widget is placed on
the screen with flow lines
showing ANDs (sequential
flow) and ORs (parallel
flows).
Summary
 The challenge now is to broaden the spectrum
of applications by
 improved user interface design
fast database search
compression methods
 It can become a general approach attached to
every database system, spreadsheet, etc.
 Research directions include:
(1) database and display algorithms
(2) user interface design.
Data Visualization Sliders
Sliders provide a threshold bar within a
scale that user can manipulate with a
mouse to select a value.
The effectiveness of sliders may be
increased by using the space inside the
sliders as
An interactive color scale
A barplot for discrete data
A density plot for continuous data
Sliders in Information Visualization
Double-edged slider with upper and
lower thresholds on network maps.
Categorical slider on software display
for selecting an arbitrary subset. (see
next slide for an example)
Ahlberg and Shneiderman’s FilmFinder
uses a suite of double-edged sliders.
Seesoft (Example of categorical slider)
Lines corresponding to service-affecting messages
(asserts, audits, peripheral interrupts, and
prms) are shown using different colors (gray levels).
Filtering
It’s the idea underlying these
applications.
Sliders are used to restrict the
information portrayed on the screen,
thereby pruning the visual clutter and
enabling the analyst to see important
underlying patterns.
Example: Frost
Improvements Upon Traditional Sliders
 The space inside the slider is used as a color
scale, thereby efficiently utilizing screen real
estate.
 The data values are shown as tick marks in a
“rug plot” in versions A and B and as the bar
lengths in version D.
 The distribution of the data is shown as a
density plot in version B and as the bar lengths
in version D.
 Selected or turned-on, regions are shown in
colors and unselected or turned-off regions are
in dark gray. (Similar to operating a paint
program)
Summary Of Important Ideas
 Enabling a user to specify an arbitrary number of
disconnected intervals while preserving the
intuitive slider interface.
 Using the space inside the slider as a color scale.
 Interactively rebinding the colors either to the
active bars or adjusting the color divisions.
 Presenting the distribution of the data.
 Showing individual data values, either as tick marks
or as bar lengths.
 Moving between the representations under user
control, enabling the users to explore from several
perspectives.
 Linking sliders to the data they control suggests
many natural and obvious extensions.