Graphics and Graphic Information Processing J. Bertin Hilary Browne

advertisement
Graphics and
Graphic Information Processing
J. Bertin
Hilary Browne
Jeff Carver
CMSC 838
September 9, 1999
Our favorite sentence
• “a problem with n characteristics is not the
sum of n problems with two characteristics”
Problem
• Collection of objects that are described by n
characteristics
• Need a way to visually represent that
information
• n = 1, 2, 3 are special cases (easy)
• n > 3 - “impassable barrier”
Data Tables
• Raw data is transformed into data tables
• x-axis - objects
– ex. movies
• y-axis - characteristics
– ex. rating, length
• Special case - networks
Object Types (x-axis)
• Reorderable
– ex. individuals
• Ordered
– ex. months
• Topographical
– ex. cities
Characteristic Types (y-axis)
• Nominal
– ex. Movie titles
• Ordinal
– ex. Movie ratings
• Quantitative
– ex. Movie length
Data table --> Graphic Constructs
• Graph choice depends on object type
– diagram
– map
– network
• Use “Synoptic” to choose
Synoptic
Easy case: n < 3
• Reorderable objects
– Reorderable matrix (bar graph)
• Ordered objects
– image-file
– array of curves
• Topographical objects
– map
Scatter Plots
• Applicable to ordered and reorderable
objects
• Useful only when n < 3
Hard case: n >3
• Useful graphic constructs
– Reorderable matrix : permutable in x and y
– Image-file : permutable in y
– Array of curves : permutable in y
• Less useful graphic constructs
– Collection of scatter plots
– Collection of maps
Synoptic
Special Case: Networks
• Reordered, Ordered, and Topographical
representations
• Can be converted to data table
Using the Synoptic
• Reordering can lead to discovery of patterns
• Deviating from suggested construction
leads to loss of information and requires
justification
• Choosing between a map and a diagram
• Size limitations
Critique
• Strengths
– Simple, Usable Taxonomy for static graphics
– Synoptic diagram
• Weaknesses
– Excerpt from book
• Lack of examples
• Terminology
– Outdated for visualization
Contributions
• Classification scheme for 2D graphical
presentation
– Used in other applications e.g. Mackinlay
• Viewing more than 3 characteristics is hard
• Using the wrong tool can lead to
information loss
Related Paper:
DeFanti - 1987 NSF report
• Scientific visualization
– “Interactive representations of scientific data”
• Established the term “visualization” in
computing
• Recommended funding the development
and use of new tools
• Automation and extension of Bertin work
• Attempt to overcome “impassable barrier”
Where has it gone?
• Automated
• Extended to 3D
• Incorporated into Dynamic Visualization
tools
– ex. Film Finder, Spotfire
Download