Graphics and Graphic Information Processing J. Bertin

1
Graphics and Graphic Information Processing J.
Bertin

• Presented by Fusun Yaman

2
Overview

• Introduction
• Description of the paper
• My favorite sentence
• Contributions
• Notes on the references
• Critique
• What happened to this topic

3
Introduction

• Section from Graphics and Graphic Information
  Processing (1977/1981)
• Problem addressed in section B
• Collection of objects that are described by n
  characteristics
• How to graphically represent this information
  when usually n gt 3

4
Terminology

• Information is in Data Table
• Objects correspond to cases (A, B, C, D)
• Characteristics correspond to variables
  (income,education, experience)

5
Terminology (continued)

• Objects can be
• Ordered (0) , like months
• Reorderable (?), like individuals
• Topographic (T), like cities
• Characteristics can be
• Nominal, like movie titles
• Ordinal, like movie ratings
• Quantitative, like length of the movie

6
Impassable barrier

• Image has only 3 dimensions
• This barrier is impassable
• Le n be number of variables (rows)
• n ? 3 Use scatter plots
• n gt 3 Other solutions needed

7
Solutions for n gt 3

• Constructing several scatter plots
• Sacrificing overall relationship
• Constructing a matrix
• Overall relationship is discovered by permutations

8
Synoptic

• Classifies graphic constructions according to two
  properties of Data Table
• If n is number of characteristics
• n gt 3 and n ? 3
• Nature of objects
• Ordered , reorderable, topographic

9

10
Graphics for n ? 3

• Matrix construction when objects are reorderable

11
Graphics for n ? 3

• Arrays of curves when objects are ordered

12
Graphics for n ? 3

• Scatter plots for both reorderable and ordered
  cases
• Third row is represented by the size of the
  marker (9)

13
Graphics for n ? 3

• In topographies bi- or tri-chromatic
  superimposition reveals the overall relation
  ships

14
Graphics for n gt 3

• Objects and characteristics are reorderable (??)
• Reorderable matrix
• Objects are ordered, characteristics are
  reorderable
• Image file (2)
• Array of curves when slops are meaningful (3)
• Ordered objects and characteristics
• Collection of tables or maps (4,5)
• Use super imposition to discover similar groups

15
Reorderable Matrix

• Objects and characteristics are reorderable (??)
• Permutable in x and y
• Overall relationship is discovered by
  permutations
• What if characteristics are not nominal?

16
Special Cases for (??)

• Weighted matrix
• Areas become meaningful
• Applicable to a data table in which row and
  column totals are meaningful
• Limited in dimension
• Matrix-file
• When one of the dimensions is too large
• Constructed similar to image files
• Use sorting to discover correlations

17
Image File

• Used for ordered objects and reorderable
  characteristics
• One card for each characteristic
• Values greater than the mean of that row are
  darkened

18
Matrix-File

• Special case for permutable matrix one of the
  dimensions is too big.
• Large number of objects across a small number of
  characteristics.
• Constructed similar to image files
• Use sorting to discover correlations

19
Matrix-File Example

• Ordered by salary, origin, age
• Higher salaries are paid to men, who are married,
  older and who have more childeren then others

20
Graphics for Networks

• A network portrays the relationships that exists
  among the elements of a single component.
• can also be represented in matrix form
• If this component is
• Reorderable network is transformable on a plane
  (19)
• Ordered network is transformable on one
  dimension (20)
• Topography non-transformable ordered network
  (21)

21
Utilization of Synoptic

• Using synoptic choose the appropriate graphic
  construction for your data
• Deviating from suggested construction leads to
  loss of information and requires justification
• Size limitations

22
My favorite Sentence

• A problem involving n rows does not correspond
  to n problems involving one row.
• Graphics is a strict and simple system of
  signs, which anyone can learn to use and which
  leads to better understanding.

23
Contributions

• Synoptic
• Classification scheme for 2D graphical
  presentation
• Permutation Matrix
• General solution for more than 3 variables
• (In the book) Identifies seven visual variables
• Position,size, value, orientation, color, texture
  and shape

24
References

• The book has no reference section!
• Semiology of graphics Diagrams, networks, maps,
  J. Bertin, 1967
• Identifies basic elements of diagrams
• Describes a framework for their design

25
Critique

• Strength of the paper
• One image summerizes his all theory on graphic
  construction selection
• Weakness of the paper
• No 3D discussion
• Not easy to follow, lack of examples (in the
  given section)
• Outdated implementation techniques

26
What happened to this topic?

• Formed a basis for research in Information
  Visualization
• Graphical constructions and ideas presented in
  this section are implemented in information
  visualization tools
• Tablelens (matrix file)
• Spotfire (scatter plots using seven visual
  variables)

27
What happened to this topic?

• Classification enabled auotomation studies
• Automating the design of graphical presentations
  of relational information, Mackinlay
• 1987 NSF report, DeFanti (uses the term
  visualization)
• Extension to 3D graphics
• Information Animation Applications in the capital
  markets, Wright
• 1987 NSF report, DeFanti