Automating the Design of Graphical Presentations of Relational Information

1
Automating the Design of Graphical Presentations
of Relational Information

• Jock MacKinlay

Beth Weinstein March 14, 2001
2
Paper Outline

• Introduction
• Expressiveness
• Effectiveness
• Composition
• Implementation

3
Introduction

• Previously application designers had to
  anticipate every situation and predesign
  graphic designs
• In related work, the BHARAT system, automatically
  creates 2D graphical designs based on the data
  type. For example, the system automatically
  chose to show the data as a line chart if the
  data was continuous.

4
Introduction

• MacKinlays goal was to build a presentation tool
  (APT) that would create a graphic design that
  would express relations and their basic
  properties effectively
• APT would have the best way to view 2-D data
  using expressiveness and effectiveness criteria.
  The paper focuses on 2-D static presentations of
  relational data (scatter plots, bar charts, etc.)
  

5
Introduction

• Three types of data
• ordinal (Monday, Tuesday, Wednesday)
• nominal (United States, Mexico, Canada)
• quantitative (1, 2, 3)
• Problems could arise when there are multiple and
  conflicting criteria for a data set

6
Expressiveness Criteria
identify graphical languages that express the
desired information

• Expressive if
• include all data in the set
• include only data from the set

7
Expressiveness Criteria
8
Effectiveness Criteria
identify which graphical languages is most
effective at using the output and human
capabilities

• Based on human perceptual capabilities
• Graphical design interpreted quickly and
  accurately
• Cleveland and McGills conjectural theory -
  different graphical designs are interpreted by
  people with varying accuracy

9
Expressiveness Criteria
10
Effectiveness Criteria
11
Composition

• Merge different encoding techniques not usually
  combined

12
Implementation

• Partitioning
• A divide and conquer algorithm
• Partition on most important element
• Selection
• For each partition, a list of graphic design is
  generated based on expressiveness criteria
• Then, the list is ordered by the effectiveness
  criteria
• Composition
• Each partitions graphic design is tested to see
  if they both can be applied, if not the next most
  effective graphic design is used

13
Favorite Sentence
an important responsibility of a user interface
is to make intelligent use of human visual
abilities and output media whenever it presents
information to the user.
14
Reference Notes

• MacKinlay - Ph.D. dissertation
• Cleveland and McGill
• General, well-cited references
• Tufte
• Knuth

15
Critique

• Strengths
• Proves that a graphical design may not be
  appropriate for all types of data
• Gives a criteria for perceptual tasks
• Weaknesses
• Vague on how extended Cleveland and McGills
  criteria
• Composition algebra unnecessary

16
Contributions

• Provides a ranking of perceptual tasks for
  nominal, ordinal, and quantitative data
• Contributes a new graphical designs by combining
  perceptual tasks
• Creates a tool that will create expressive and
  effective graphical designs based on the data

17
(No Transcript)