Transcript Dynamic Queries for Visual Information Seeking

Dynamic Queries
–presented by Bhaskar Chatterjee

Visual Alternative to SQL for Querying
databases



Depending on data types and the values
decides the input controls for the search
Examples are Buttons, Ratio Buttons,
Simple sliders and Range Sliders etc
Two papers: deals with the rationale for
dynamic queries and also go over two
Applications: Home Finder, Film Finder
Dynamic Queries

Constantly need to update queries as
users adjust sliders or select buttons




Visual representation of query
Visual representation of results
Rapid and incremental control of query
Immediate and continuous feedback
Advantages



Users can “fly through data” by adjusting
sliders
Novice: formulating query at command line
leads to errors in syntax and understanding
Experts: interpretation of results can be
easier-air traffic controllers, demographers,
statisticians
Examples: Geographic applications
Response of 18 Subjects using
HomeFinder
Example: Cancer Rates in US
Example (Education Application)
Alphanumeric tabular display:
directory display in UNIX


This lets users explore UNIX directory
by size and age
Three versions developed



Highlighting matches with color
Highlighting matches with asterisk
Displaying only the matching lines and
hiding the rest
Disadvantages observed

The dynamic query is poorly matched with
the current hardware and software systems




Current database tools cannot satisfy the
requirements for rapid searches
Application specific programming is required to
take advantage of the dynamic query methods.
Standardized input output software are required to
integrate to existing system
Visually handicapped find it more difficult
Research Directions




Database and display algorithms
 To handle rapid display updates
Special data structures kept in rapid storage
can handle rapid algorithms
Data organized in buckets, for example 1 to
50000 can be organized in 100 buckets
(each having 500 data points)
Use of parallel hardware and algorithms,
data compression to store large amounts of
data
Research (Contd.)

User interface design


People should be able to grasp the content of the
picture
Questions:



How to allow user to use Boolean combination of
slider settings
Choose among highlighting among color, points of
light etc.
Allow varying degrees of intensity, large
points/zooming
Other Research Issues

Display Issues:



What color, size and shape to use ?
How to represent overlapping data on a map?
Input Issues:



Widget design is a central issue:
Limits and display of Current Values, number of
boxes
Multidimensional input ( currently they are
clumsy) – research is in 3D mouse: 6 degree of
freedom.
Examples of 2-Dimensional
input
Visual Information Seeking: Tight
coupling with dynamic queries and
Starfield Display



Dynamic Queries Filter: query parameters
rapidly adjusted with slider, buttons etc.
Starfield Display: result sets are
continuously available and support viewing of
hundreds or thousands of items
Tight Coupling: query components are
interrelated in ways that preserve display
invariants and support progressive
refinement.
Tight Coupling



Tight coupling reveals the software state and
constrains the user from making erroneous
actions
For example: if a user wants films before
1935 then only certain actors and directors
are further selectable.
Tight coupling aspect: every output of query
is a candidate for input of a another query

Helps in reducing screen clutter
Tight Coupling (Contd.)



Other Aspects :
Progressive refinement of query
Details on demand : idea of hypermedia

Click on the data points to get further
information
FilmFinder
FilmFinder (Contd.)
FilmFinder (Contd.)
FilmFinder (Contd.)