Graph and Network Visualization

8/7/2019 Graph and Network Visualization

1/69

Graph and NetworkVisualization

John Stasko

Georgia Institute of Technology


2/69

InfoVis 2

Connections

Connections throughout our lives and the

world Circle of friends

Deltas flight plans

Model connected set as a Graph


3/69

InfoVis 3

What is a Graph?

Vertices (nodes)

connected by Edges (links)

1 2 30 1 0

1 0 10 1 0

1

23

1: 22: 1, 33: 2 1

32

Adjacency matrix

Adjacency list

Drawing


4/69

InfoVis 4

Graph Terminology

Graphs can have cycles

Graph edges can be directedorundirected

The degreeof a vertex is the number ofedges connected to it

In-degreeand out-degreefor directed graphs

Graph edges can have values (weights)on them (nominal, ordinal or quantitative)


5/69

InfoVis 5

Trees are Different

Subcase of general graph

No cycles Typically directed edges

Special designated root vertex


6/69

InfoVis 6

Graph Uses

In information visualization, any number

of data sets can be modeled as a graph US telephone system

World Wide Web

Distribution network for on-line retailer

Call graph of a large software system

Semantic map in an AI algorithm Set of connected friends


7/69

InfoVis 7

Graph Visualization Problems

Graph layout and positioning

Make a concrete rendering of abstract graph

Scale

Not too much of a problem for small graphs,but large ones are much tougher

Navigation/Interaction

How to support user changing focus andmoving around the graph


8/69

InfoVis 8

Layout Algorithms

Entire research communitys focus

Good references: Tutorial (talk slides)

www.cs.brown.edu/people/rt/papers/gd-tutorial/gd-constraints.pdf

G. diBattista, P. Eades, R. Tamassia, and I.Tollis, Graph Drawing: Algorithms for theVisualization of Graphs, Prentice Hall, 1999.


9/69

InfoVis 9

General GD Information

Good web links www.cs.brown.edu/people/rt/gd.html

www.research.att.com/sw/tools/graphviz/

rw4.cs.uni-sb.de/users/sander/html/gstools.html


10/69

InfoVis 10

Graph Drawing Conference


11/69

InfoVis 11

Vertex Issues

Shape

Color

Size

Location Label


12/69

InfoVis 12

Edge Issues

Color

Size Label

Form Polyline, straight line, orthogonal, grid,

curved, planar, upward/downward, ...


13/69

InfoVis 13

Aesthetic Considerations

CrossingsCrossings-- minimize towards planar

Total Edge LengthTotal Edge Length-- minimize towards properscale

AreaArea-- minimize towards efficiency

Maximum Edge LengthMaximum Edge Length-- minimize longestedge

Uniform Edge LengthsUniform Edge Lengths-- minimize variances

Total BendsTotal Bends-- minimize orthogonal towardsstraight-line


14/69

InfoVis 14

Which Matters?

Various studies examined which of the

aesthetic factors matter most and/or whatkinds of layout/vis techniques look best Purchase, Graph Drawing 97

Ware et al, Info Vis1(2), June 02

Ghoniem et al, Info Vis 4(2), Summer 05

Results mixed: Edge crossings seemimportant


15/69

InfoVis 15

Layout Heuristics

Layout algorithms can be

planar

grid-based

orthogonal

curved lines

hierarchies

circular ...


16/69

InfoVis 16

Types of Layout Algorithms

From:P. Mutzel, et al

Graph Drawing 97


17/69

InfoVis 17

Layout Examples

Cool java applethttp://java.sun.com/applets/jdk/1.2/demo/applets/GraphLayout/example1.html

Examples of dynamic graph layoutalgorithms


18/69

InfoVis 18

Scale Challenge

May run out of space for vertices and

edges (turns into ball of string) Can really slow down algorithm

Often use clusteringto help

Extract highly connected sets of vertices

Collapse some vertices together


19/69

InfoVis 19

Navigation/Interaction Issues

How do we allow a user to query, visit, or

move around a graph? Changing focus may entail a different

rendering


20/69

InfoVis 20

Graph/Network Visualization

One of the oldest and most studied areas

of information visualization

Good overview

Herman et al, IEEE TVCG 00


21/69

InfoVis 21

Examples www.visualcomplexity.com


22/69

InfoVis 22

Graph Uses

Facilitate understanding of complex socio-

economic patterns Social Science visualization gallery (Lothar

Krempel):

http://www.mpi-fg-koeln.mpg.de/~lk/netvis.html

Next slides: Krempel & Plumpers studyof World Trade between OECD countries,1981 and 1992


23/69

1981http://www.mpi-fg-koeln.mpg.de/~lk/netvis/trade/WorldTrade.html


24/69

1992


25/69

InfoVis 25

Social Network Visualization

Social Network Analysis (Linton Freeman)

http://www.sfu.ca/~insna


26/69

InfoVis 26

People connections

Charles Isbell, Cobot


27/69

InfoVis 27

Vizster

Saw in Social Visualization lecture


28/69

InfoVis 28

SocialAction

Perer & Shneiderman

InfoVis 06


29/69

InfoVis 29

Graph Uses

Facilitate understanding of network flows,

relations

Even information with a geographical

content can best appear as a network railmaps


30/69


31/69


32/69


33/69

InfoVis 33

3 Subway Diagrams

Geographic landmarks largely suppressed

on maps, except water (rivers in Paris,London) and asphalt (highways in Atlanta)

Rather fitting, no?

These are more graphsgraphsthan maps!


34/69

InfoVis 34

Case Study

SeeNet

Visualizing network data (phone traffic)R. Becker, S. Eick and A. WilksAT&T


35/69

InfoVis 35

Domain

AT&T long distance phone network

110 Nodes (switches)Geographical location

Connected by 12,000 links

Directed, almost completely connected

Data every 5 minutes

EARTHQUAKE!!!EARTHQUAKE!!!

Oct. 17, 1989


36/69

InfoVis 36

Questions

Where are the overloads?

Which links are carrying most traffic? Was there network damage?

Is there underutilized capacity?

Are calls getting in to affected area or arethere bottlenecks?

Is overload increasing or decreasing?


37/69

InfoVis 37

Edge Drawing Strategies

116Label

Thickness

Color

116

29Directed


38/69

InfoVis 38

Problems

Too many lines!

Occlusion

Long lines become more important

Cant see what happens in Midwest

Solutions Use half/half technique out/out

Draw most important last Use thickness & color for traffic


39/69

Earthquake data


40/69

InfoVis 40

More Help

Shorten all lines so as to de-emphasize

transcontinental links


41/69


42/69

InfoVis 42

Case Study

NicheWorks

Interactive Visualization of Very Large GraphsGraham WillsLucent (at time)


43/69

InfoVis 43

Big Graphs

20,000 - 1,000,000 Nodes

Works well with 50,000 Projects

Software Engineering

Web site analysis

Large database correlation

Telephone fraud detection


44/69

InfoVis 44

Features

Typical interactive operations

Sophisticated graph layout algorithm 3 Layouts

Circular

Hexagonal

Tree

3 Incremental AlgorithmsSteepest Descent

Swapping

Repelling


45/69

InfoVis 45

Web Site Example

Circle layout Hexagonal layout Tree layout

I t f


46/69

InfoVis 46

Interface

I t f


47/69

InfoVis 47

Interface

Ph F d E l


48/69

InfoVis 48

Phone Fraud Example

40,000 calls35,000 callers

Shown arepeople callingthat country

Length of

edge isduration ofcall

F d E l


49/69

InfoVis 49

Fraud Example

Filtering for peoplewho made multiplecalls and spent asignificant amountof time on the phone

Playing with parameters

like these is importantbecause fraudstersknow how to evade

Note the two peoplecalling Israel and Jordan

F d E l


50/69

InfoVis 50

Fraud Example

Zooming in, we notice they havesimilar calling patterns and numbers

(likely part of same operation)

Illegal to call between Israel andJordan at the time, so fraudsters

set up rented apts in US and chargeIsraeli and Jordanian business peoplefor 3rd party calling

When bills came to US, they would

ignore and move on

M N t St ff


51/69

InfoVis 51

More Neat Stuff

http://willsfamily.org/gwills/

Lots of interesting application areas More details on NicheWorks

Oth A li ti


52/69

InfoVis 52

Other Applications

Email

How would you visualize all email traffic inyour department between pairs of people?

Solutions???

Sol tions


53/69

InfoVis 53

Solutions

Put everyone on circle, lines between

Color or thicken line to indicate magnitude Use spring/tension model

People who send a lot to each other are

drawn close together Shows clusters of communications

More Email


54/69

InfoVis 54

More Email

How about visualizing internet traffic?

Byte traffic into the ANS/NSFnet T3 backbone for the month of November, 1993


55/69

http://www.ncsa.uiuc.edu/SCMS/DigLib/text/technology/Visualization-Study-NSFNET-Cox.html

Inbound traffic measured in billions of bytes on the NSFNET T1 backbone for September 1991


56/69

Linux kernel


57/69

InfoVis 57

Linux kernel

http://perso.wanadoo.fr/pascal.brisset/kernel3d/kernel3d.html

TouchGraph


58/69

InfoVis 58

TouchGraph

www.touchgraph.com

Focus of Graph


59/69

InfoVis 59

Focus of Graph

Particular node may be focus, oftenplaced in center for circular layout

How does one build an interactive systemthat allows changes in focus?

Use animation

Intuition about changes not always right

Focus Change Animation


60/69

InfoVis 60

Focus Change Animation

Straight linear interpolation

of focus changes not asappealing as changes alongpolar coordinates

Yee, Fisher, Dhamija, HearstInfoVis 01

Video

MoireGraphs


61/69

InfoVis 61

MoireGraphs

Combine a focus + context radial graphlayout with a variety of interactiontechniques

Useful when visual information such as

images is also present at each node andmust be displayed

Jankun-Kelly & MaInfoVis 03

Navigation and interaction


62/69

InfoVis 62

Navigation and interaction

Video

Alternative Approaches


63/69

InfoVis 63

Alternative Approaches

Recently, researchers have been exploringapproaches other than the traditionalnode-link views

Adjacency Matrices


64/69

InfoVis 64

Adjacency Matrices

MatrixExplorer

Henry & Fekete

InfoVis 06

PivotGraph


65/69

InfoVis 65

PivotGraph

Cluster nodes on an attribute valueShow links between sets Wattenberg

CHI 06

Semantic Substrates


66/69

InfoVis 66

Semantic Substrates

Shneiderman & Aris

InfoVis 06

Separate nodes by attributeUse dynamic querycapabilities

More Resources


67/69

InfoVis 67

Network visualization resources http://www.caida.org/projects/internetatlas/viz/

Good article on graph layout http://www.csi.uottawa.ca/ordal/papers/sander/main.html

More to Come...


68/69

InfoVis 68

Topic of WWW/InfoSphere (next lecture)will touch on graphs and networks too

Lots of example visualizations

End


69/69

InfoVis 69

References

Spence and CMS texts

All referred to papers and web sites

Dagon and Leahy, F 99 slides

Documents

Graph and Network Visualization