Post on 17-Dec-2015
INFORMATION MURALA technique for displaying and
navigating large information spaces
Dean F. Jerding and John T. StaskoGraphics, Visualization, and Usability Center
College of ComputingGeorgia Institute of Technology
MOTIVATION
Global visualizations for large data sets are constrained by the limited number of pixels on the screen
scalability problem addressed
WHAT IS INFORMATION MURAL ?
Information mural is a
– 2D, miniature representation– representing an entire information space– using visual attributes such as color and
intensity to portray information density
OBJECTIVES
Create a representation of a large information space on the screen
Mimic the original visual representation
Minimize the loss of information in the compressed view
AUTOMOBILE DATA
Standard parallel coordinate view of the data
AUTOMOBILE DATA
Mural of a parallel coordinate view of automobile data showing MPG, engine displacement, horsepower, weight, acceleration, and model year (1970 – 1982)
AUTOMOBILE DATA
Color overlaid for number of cylinders (3 = red, 4 = orange, 5 = yellow, 6 = green, 8 = cyan).
THE TECHNIQUE
First present the global view to the user
– Scale the original visualization– Make explicit the density of overlap of
information visually apparent (using colors)
Allow user to analyze or navigate the global view for detailed views on the items of interest
INFORMATION DENSITY REPRESENTATION
This is the distinguishing feature of this paper.
Problem – Many pixels on the original image contribute to a
single pixel in the mural– Each pixel in the mural can have a single color
Constraints– Cannot resort to using the RGB components
INFORMATION DENSITY REPRESENTATION
Some earlier approaches are – Aggregation– Sampling
SolutionColor the pixel according to the attribute
color that occurs most frequently in that pixel
THE BIGGER PICTURE
MURAL
ORIGINAL IMAGE
INTENSITY MAPPING(cont)
1. Construct a unit square connecting the centers of the surrounding four destination pixels
INTENSITY MAPPING (cont)
2. Using the scaled location of the center of the original pixel to divide the square into four quadrants
INTENSITY MAPPING (cont)
3. The area of the quadrant diagonally opposite to each of the four destination pixels is the amount of intensity contributed by that pixel
SUITABLE FOR ?
Navigating information spaces
Studying the characteristics of geographic/spatial data
Discovering attributes and relationships
A CLOSER LOOK……
Parallel program message trace diagram drawn by just overplotting
Mural of parallel program message trace
SUN SPOT MURAL
Mural of the number of sun spots recorded daily, 1850-1993.
SUN SPOT MURAL
Mural of the number of sun spots recorded daily, 1850-1993,
zoomed in on a small area
SUN SPOT MURAL
View of sun spots showing focus area and mural of entire data
set at the bottom.
DOCUMENT VISUALIZATION
DOCUMENT VISUALIZATION
Murals showing keyword distribution for search on ``visualization'' (yellow), ``object-oriented'' (green), and ``OO'' (cyan) in three documents
OPINIONS A new approach to representing of data
density Examines all the points in the source image –
may be slower For very large data mural becomes fuzzy Colors - double edged sword
– helps in identifying strong patterns– noticing the colors – appropriateness of colors
HCI METRICS
User satisfaction - good Learning time - good User Performance } depends Error rate } on Retention } visualization
Thank you