Visualization Concepts: Overview

Current exploitation of information accessible by computer: a fraction ! Future increase of data rates expected

• increased access to network information
•  
• increase of computer power
•  
• increase of output from measuring devices
•  

=> Need for systematic strategies (concepts, methodologies, intelligent visualization systems) to exploit data [ROB94]

Two strategies

• "Grand Tour of Visualizations"
  
  • Example: 4 data variables, 4 visual attributes available
    e.g. (leaf length, leaf width, leaf type, leaf age)
    => (position x, position y, symbol type, symbol size)
    offers 24 possibilities!
  • General rule of thumb
    n data characteristics, m visual attributes (n>m): n!/(n-m)!
  •  
• "Renaissance teams" (Donna Cox, NCSA)
  Domain expert + visualization expert create visualizations

Use of mapping constraints

Arising from

• data (data characteristics)
•  
• reality (problem domain)
•  
• computer environment (available software and hardware)
•  
• viewer (interpretation aims = visualization aims)
•  
• viewer (abilities and desires of user)
•  
• viewer (generation of "meaningful" pictures)
•  

Data characteristics

Data characteristics include

• number and type of dimensions (e.g. 1,2,3; spatial, spectral, temporal)
•  
• relations between data variables
•  
• data type (ordinal, nominal, quantitative)
•  
• data reliability, accuracy
•  

Interpretation aims

Interpretation aims are defined by the viewer(s), e.g., for

• Scientific visualization, e.g.
  
  • identify objects, compare values, distinguish objects, categorize objects
•  
• Software visualization, e.g.
  
  • focus on text/data structures/performance/algorithm
•  
• Information visualization, e.g.
  
  • focus on detail with overall view, view relations

Abilities and desires of user

Restrictions by

• Color perception ability
•  
• Color memory
•  
• Intuitive color ranking
•  
• Mental rotation ability
•  
• Fine motor coordination ability
•  
• Preferences
•  

[DOM94]

Available software and hardware

Restrictions by

•  
• Output devices (e.g., black/white monitor)
•  
• Input devices (e.g., mechanical/optical mouse, keyboard)
•  
• Lack of computation/graphics power: e.g, real-time performance constraints
•  
• Software available (e.g., volume visualization)
•  

"Meaningful pictures"

Coherent visual representations

• Bottom-up approaches [MAC86], [SEN94]
  
  • generate pictures from visual primitives or graphical language
  • test pictures for expressiveness and effectiveness
• Top-down approaches [WEH90], [ROB91]
• • allow only coherent (complex) techniques
  • choose best fitting technique

Use of appropriate visual attributes (visual cues)

Approaches to systematic strategies: Overview

• Mackinlay (APT)
•  
• Roth and Mattis (SAGE)
•  
• Casner (BOZ)
•  
• Senay and Ignatius (VISTA)
•  
• Robertson (NSP)
•  
• Wehrend and Lewis (Catalog of Visualization Techniques)
•  
• Haber and McNabb (Visualization Idioms)
•  
• Beshers and Feiner (AutoVisual)
•  
• Robertson, Card and Mackinlay (Information Visualizer)
•  

Selected concepts of visualization systems

Mackinlay (APT)

A Presentation Tool [MAC86]

Automatic 2-d discrete data presentation of relational information

• graphics primitives: e.g. areas, lines, marks
•  
• visual attributes: e.g. color, size, saturation
•  
• primitive graphical languages and composition rules ß complete graphs
•  
• testing result: expressiveness and effectiveness criteria must hold ("generate-and-test")
•  

Roth and Mattis (SAGE)

SAGE [ROT90]

• Knowledge-based system to create 2-d graphics automatically
•  
• Elaborate characterization of data, semantics and relations
•  
• total-costs = material - costs + labor-costs, reflects composition in graph (e.g. two bar segments for each represented interval)
•  

Includes components for constructive design of graphics (SageBrush) and retrieval of graphics (SageBook).

Casner (BOZ)

[CAS91]

Approach from task analysis

Operating on relational database to produce 2-d graphics

• first criteria is TASK: describe task through logical operators
•  
• map logical operators to perceptual operators
  (search operators/computation operators)
  =>design graphs to satisfy visualization tasks
•  
• optimal graph has lowest cost for performing task
•  

Senay and Ignatius (VISTA)

VISualization Tool Assistant: extension to 3d visualizations [SEN94]

Knowledge-based system to automatically design visualizations

• primitive graphics techniques are composed to complex visualizations
•  
• extended composition rules for 3-d: transparency, occlusion, intersection extensive rules on effective mapping strategies
•  
• emphasize preattentive, if feature should attract viewer's attention
•  
• user may modify design interactively
•  

Robertson (NSP)

Natural Scene Paradigm [ROB91]

• data elements are mapped onto features of natural scenes
  
  • e.g. mountains and valleys, showing patterns of density and color
•  
• constraints
  
  • data characteristics:
    dimensionality, relationships, ordinal/nominal, discrete/continuous
  • interpretation aims:
    point/local/global importance; relative importance
  • display constraints (user directives)

=>assures coherency through top-down design of complex scenes

=>assures problem-free interpretation through perceptual skills of humans

Wehrend and Lewis (Catalog of Visualizations)

Classification of simple and complex visualization techniques [WEH90]

Categorize each visualization technique by:

• what kind of data can be displayed ("attributes")
  
  • attributes: [scalar, scalar field, nominal, direction, direction field, shape, position, spatially extended region or object, structure]
• what operations act on these attributes ("operations/judgments").
  
  • operations: [identify, locate, distinguish, categorize, cluster, distribution, rank, compare within and between relations, associate, correlate]

"Catalog of visualization techniques": large 2-d matrix to identify meaningful visualization techniques for a pair of (attribute/operation).

Haber and McNabb

Visualization process is series of transformations to convert raw simulated data into a displayable image:

Visualization idiom: "a specific sequence of data enrichment and enhancement transformations, visualization mappings and rendering transformations that produce an abstract display of a scientific data set".

Beshers and Feiner (AutoVisual)

Rule-based design of interactive multivariate visualizations (n-Vision) [BES93]

• Specify visualization tasks
  • task operators: fundamental cognitive operations (exploration, directed search, comparison)
  • task selections: subset of relations, data items to display (constraints on available variables)
• Use rule-based visualization design principles
  • taking into account: visualization task, capabilities of hardware, characteristics of data
  • using interactive Worlds within Worlds techniques
• Termination conditions
  • potential expressiveness
    (visualization has potential, under user control, to display all its assigned information over time)
  • potential effectiveness
    (visualization has potential, under user control, to display its assigned information sufficiently clearly over time")

Robertson, Card and Mackinlay (Information Visualizer)

Paradigm to optimize cost structure for finding and accessing information [ROB93]

• Common data characteristics: information organized in data bases
•  Common problems: Costs for
  • retrieval of information from distant source
  • accessing that information once in use
•  Solution
  • Information workspaces

Information workspaces characterized by

• use of (virtually) large workspaces -- virtual large screen space; increase density of information (perspective, 3-d)
•  use of semiautonomous agents to off-load work of users: organize information into clusters or conduct searches.
•  support of real-time interaction rates: require certain classes of object to occur at set rates, e.g. animation;
•  use of visual abstractions: use visualization of different abstract information structures, including linear or hierarchical structures, continuous or spatial data.

Sample visualization techniques: Cone trees, Perspective Wall, 3D/Rooms

Visualization Concepts
HyperVis Table of Contents

Last modified on February 11, 1999, G. Scott Owen, owen@siggraph.org