information visualization

Information Visualization

UI lab. 이 석 재

Data

Data transfer

Insight(learning, knowledge extraction)

Goal

Data

Visualization

Map: data → visual

Map-1 visual → data insight

Data transfer

Insight

Visual transfer

(communication bandwidth)

Method

Data

Visualization

Map: data → visual

Visual Mappings must be:• Computable (math)

visual = f(data)

• Comprehensible (invertible)data = f-1(visual)

• Creative!

Visual Mappings

Effectiveness

• User learnability: Learning time Retention time

• User performance: *** Performance time Success rates Error rates, recovery Clicks, actions

•User satisfaction: Surveys

Introduction

• To understand something is called “seeing” it.• Visual metaphors – a nexus of relationships between what

we see and what we think.• How have we increased memory, thought, and reasoning? -By the invention of external aids.

Multiplication Aids

• Visual and manipulative use of the external world amplifies cognitive performance.

• Why does using pencil and paper make such a difference? - What is hard is holding the partial results in memory until

they can be used. ->The visual representation, by holding partial results outside

the mind, extends a person’s working memory.

Multiplication Aids

• slider rule 1. an analogue interactive visual device that represents

quantities as scales with length proportional to their logarithms 2. actually does the visual computation

Navigation charts

• The map is not just a calculator, but also a storage device, storing for access enormous amounts of information naturally located near where they are needed for calculation.

• Diagrams can lead to great insight, but they can also lead to the lack of same.

• The decision depended on whether the temperature would make the O-rings that sealed the sections of the booster rockets unsafe.

Diagrams

Diagrams

Diagrams

INFORMATION VISUALIZATION

• VISUALIZATION Definition : The use of computer-supported, interactive,

visual representations of data to amplify cognition.

• Purpose : insight, not pictures

• Both of these visualizations show abstractions, but the abstractions are based on physical space=>SCIENTIFIC VISUALIZATION

ORIGINS OF INFORMATION VISUALIZATION

• Work in data graphics dates from about the time of Playfair(1750), who seems to be among the earliest to use abstract visual properties such as line and area to represent data visually (Tufte, 1983)

• Tukey (1977) began a movement from within statistics with his work on Exploratory Data Analysis. (Box plot)

• The first use of the term information visualization to our knowledge was in Robertson, 1989.

Active Diagrams

• The Periodic Table, originally developed by Mendeleyev, is an important diagrams in the development of chemistry.

• Figure 1.12 shows an information visualization based on the Periodic Table (Ahlberg, 1992) The user can set sliders that control which of the elements in the table will be highlighted.

LARGE-SCALE DATA MONITORING

• Information visualization to monitor and make sense of large amounts of dynamic, real-time data (decision-support application)

Information Chromatography

Visualization is used to detect telephone fraud

Information chromatography :

Patterns in the data are revealed by laying them out on a particular visual substrate.

Knowledge Crystallization

We have said that the purpose of information visualization is to use perception to amplify cognition

Knowledge Crystallization

Visualization Levels of Use

• Visualization on four levels of use(1) Visualization of the infosphere(2) Visualization of an information workspace

Visualization Levels of Use

• Visualization on four levels of use(3) Visual knowledge tools(4) Visual Objects

Cost Structure

Cost Structure

– Cost-of-Knowledge Characteristic Function

How Visualization Amplifies Cognition

(1) By grouping.(2) About a single element.(3) Easy for human

Mapping Data to Visual Form

• We can think of visualization as adjustable mapping from data to visual form to the human perceiver.

Data Table

• The usual strategy is to transform this data into relation that are mare structured and thus easier to map to visual form.

• Mathematical treatment omits descriptive information that is important for visualization

DATA TABLE

(case by variables arrays)

• Bertin(1977/1981)- cases -> objects- variables -> characteristics- function, input variable, output variable

Data Table

• Data table can undergo data transformations that affect their structure.

Data Table

• Data table can describe hierarchical and network data.

Data Table

• N = nominal variableO = ordinal variableQ = quantitative variable

• Elementary choices for data transformations derive from the variables types(Q->O, O->N, N->O)

• Subtypes that represent important properties(Qt = Quantitative Time)

Variable Type

• Metadata is descriptive information about data.• Metadata can be important in choosing visualization• An important form of metadata is the structure of a Data

Table.• Additional metadata could be explicitly to the Data Table by

adding

Metadata

Data Transformations

• Concatenated to form chains of aggregation and classing as part of the knowledge crystallization

• Can be used to detect more patterns

Data Transformations

Aggregation cycle

Data Transformations

Visual Structures

• Data tables Visual Structures Mapped with Marks and Graphical properties

• Effectiveness

Data Transformations

Visual Stuctures

Level of the visual system

1st level : Retina

- Retina is good at detecting movement or other changes

2nd level : foveola ( 황반 )

- preattentive and stereoscopic processing

3rd level : within the foveola

- 황반의 중심부에 움푹 패인 부분 ( 중심와 )

Perception

Visual information processing

Controlled processing : Textual description

- detailed, serial, low capacity, slow

Automatic processing : Poping out during search

- Parallel, high capacity, fast

Interaction among the visual codings of information

- Produce patterns

Visual Stuctures

The most fundamental aspect of a Visual Structure is its

use of space

: Spatial position is a good visual coding of data

Visual Stuctures

Spatial Substrate

Several techniques to increase the amount of

information

Composition : orthogonal placement of axes, creating a 2D

space

Alignment : Repetition of an axes at a different position in

the space

Visual Stuctures

Folding : continuation of an axis in an orthogonal dimension

Visual Stuctures

Recursion and Overloading

Visual Stuctures

View structureView structure

Connection and Encloser

1. Connection

Link connection

2. Encloser

Link encloser

Retinal Properties

- retina properties

- 자동적으로 process 되는 visual feature

- relative effectiveness of different retinal properties

Temporal encoding

- Some variable time

View structureView structure

Location Probe

View TransformationView Transformation

Viewpoint control - zoom, pan - overview + detail

Distortion - focus + context view - bifcoal lens

Distortion - focus + context view - bifocal lens

View TransformationView Transformation

Interaction and transformation controls

• Hyperbolic tree 에서 node 는 마우스를 이용하여 display의 중앙으로 드래그 할 수 있음 .