Robert Kosara1/32

InfoVis 2001Robert Kosara, Tom Theußl

Robert Kosara2/32

Conference

• Part of IEEE Visualization 2001• Paradise Point Resort, San Diego, CA• Chair: Keith Andrews• About 120-130 participants• More than PVG (100-110, according to

Keith Andrews, anyway ;-)• Interesting People

Robert Kosara3/32

Overview (1/2)

• Invited Talk: Pat Hanrahan, "To Draw ATree"

• Papers 1: Time, Change, and Clustering• TN 1: Software Testing and Focus Zooming• Papers 2 & TN 2: Graph Visualization• Papers 3: Hierarchical Visualization• Social Event ;-)

Robert Kosara4/32

Overview (2/2)

• Papers 4: Blurring, Queries, and Bar Charts• Papers 5: Empirical Studies• Case Studies 1: Decision Trees and

Clickstreams• Case Studies 2: GIS and Medicine• Capstone Panel: "Battlefield Visualization"

Robert Kosara5/32

Invited Talk: Pat Hanrahan

• "To Draw A Tree"• Very Interesting• Lots of Trees• Historical Examples

• Slides available at http://www.infovis.org

Robert Kosara6/32

P1: Time, Change, Clustering

• Marc Weber, Marc Alexa, Wolfgang Müller,„Visualizing Time-Series on Spirals“

• Lucy Nowell, Elizabeth Hetzler, Ted Tanasse,„Change Blindness in InformationVisualization: A Case Study“

• George S. Davidson, Brian N. Wylie, Kevin W.Boyack, „Cluster Stability and the Use of Noisein Interpretation of Clustering“

Robert Kosara7/32

TN1: SW Testing, Focus Zoom

• J. Eagan, M.J. Harrold, J.A. Jones, J.StaskoVisually Encoding Program TestInformation to Find Faults in Software

• T.A. KeaheyGetting Along: Composition ofVisualization Paradigms

Robert Kosara8/32

Eagen et al.: Program Testing

• Display Code• Color Coded by Test Results• Color Code Green -> Red• Seamless Zoom• Looks very useful

Robert Kosara9/32

Keahey: Getting Along

• The old "generalized F+C" stuff• Nice, smooth Pan&Zoom• Nothing new

Robert Kosara10/32

P2 & TN 2: Graph Visualization• K.-P. Yee, D. Fisher, R. Dhamija, M. Hearst

Animated Exploration of DynamicGraphs with Radial Layout

• J.M. Six, I.G. TollisEffective Graph Visualization via NodeGrouping

• F. v. Ham, H. v. de Wetering, J.J. van Wijk:Visualization of State Transition Graphs

• J. Abello, I. Finocchi, J. KornGraph Sketches

Robert Kosara11/32

Yee &al: Animated Exploration

• Tree Layout on concentric circles• Nicely animated

– Interpolation along circles– slow-in, slow-out

• Nothing really new

Robert Kosara12/32

Six &al: Effective Graph Visu.

Robert Kosara13/32

Van Ham &al: State Trans. Gr.

• Clustering nodes• 3D layout, similar to cone trees• similar groups look similar• "corn trees"

Robert Kosara14/32

Abello &al: Graph Sketches

• Huge graphs, many nodes & edges• many nodes between two edges (calls)• show adjancy matrix• zoomable• nice• (Technical Note)

Robert Kosara15/32

Papers 3: Hierarchical Vis

• Ben Shneiderman, Martin Wattenberg, DowJones, „Ordered Treemap Layouts“

• Raimund Dachselt, Jürgen Ebert, „CollapsibleCylindrical Trees: A Fast HierarchicalNavigation Technique“

• Ernst Kleiberg, Huub van de Wetering, Jarke J.van Wijk, „Botanical Visualization of HugeHierarchies“

Robert Kosara16/32

Papers 3: Hierarchical Vis

• Ben Shneiderman, Martin Wattenberg, DowJones, „Ordered Treemap Layouts“

• Raimund Dachselt, Jürgen Ebert, „CollapsibleCylindrical Trees: A Fast HierarchicalNavigation Technique“

• Ernst Kleiberg, Huub van de Wetering, Jarke J.van Wijk, „Botanical Visualization of HugeHierarchies“

Robert Kosara17/32

P4: Blur, Queries, Bar Charts

• R. Kosara, S. Miksch, H. HauserSemantic Depth of Field

• S. Havre et al.Interactive Visualization of MultipleQuery Results

• D. Keim et al.Pixel Bar Charts: A New Technique forVisualizing Large Multi-Attribute DataSets Without Aggregation

Robert Kosara18/32

Kosara &al: SDOF

• First presentation, at 8:30 in the morning• ~100 people there• good presentation ;-)• Lots of questions (chair had to cut off)• Lots of questions & praise ;-) afterwards• Indication of work on SDOF by others

Robert Kosara19/32

Havre &al: Query Results

• Query results from search engines• Gives idea of number and distribution of

results• Combine several queries

Robert Kosara20/32

Keim &al: Pixel Bar Charts

• Extension of bar charts• Pixel-based stuff, who would've thought? ;-)• Puts more information into bar charts• Correlations easier to see• Complex Layout algorithm, works most of

the time (NP-complete problem)

Robert Kosara21/32

P 5: Empirical Studies

• A. KobsaAn Empirical Comparison of ThreeCommercial InfoVis Systems

• T. Barlow, P. NevilleA Comparison of 2-D Visualizations ofHierarchies

• M. Tavanti, M. Lind2D vs. 3D, Implications on SpatialMemory

Robert Kosara22/32

Kobsa: Comparison

• Eureka (TableLens), InfoZoom, Spotfire• 83 participants! (students, "experts in dating

and cars")• Some significant differences (for details see

paper)• Discussion if selection of systems and

tasks was good

Robert Kosara23/32

Barlow &al: Hierarchies

• Four different tree visualizations• 15 participants• No very strong results

• Taped presentation, still over time

Robert Kosara24/32

Tavanti &al: 2D vs. 3D

• Simple memory test• Simple layouts, no

overlap in 3D• 20 subjects• two-stage experiment• 3D better for this task• A bit disappointing

Robert Kosara25/32

C1: Decision Trees, Clckstrms

• T. Barlow, P. NevilleVisualization for Decision Tree Analysisin Data Mining

• J. Brainerd, B. BeckerE-Commerce Clickstream Visualization

Robert Kosara26/32

Barlow &al: Decision Trees

• Multiple, Linked views• Different styles: Trees, Aggregation, ...• Zoomable• Interface for Data Mining: Training,

Validation, Comparison

Robert Kosara27/32

Brainerd &al: Clickstreams

• Website analysis ("E-Commerce")• Shows user behaviour• Aggregation, Differentiation by type• Looks very effective

Robert Kosara28/32

C2: GIS and Medicine

• R.M. Edsall, A.M. MacEachren, L. PickleDesign and Assessment of an EnhancedGeographic Information System forExploration of Multi-Variate HealthStatistics

• J. Agutter et al.Graphic Data Visualzation forCardiovascular System

Robert Kosara29/32

Edsall &al: GIS and Medicin

• Linked views: Parallel coordinates, scatter,etc.

• High-dimensional data

• Looks useful

Robert Kosara30/32

Agutter &al: Cardiovascular

• Cardiovascular data for Anaesthesia• Based on pipe metaphor• very intuitive• Tested with physicians,

no great success ...

• Looks nice, but is it needed?

Robert Kosara31/32

Capstone: Battlefield Vis

• Visualization of battlefield info• Mostly asked for help in developing stuff• People made point of leaving• Pathetic image of US armed forces

– cheesy demo "movie"– "brightest and best" ... yeah right ....

• Very bad idea, imho

Robert Kosara32/32

Conclusions

• Interesting conference!• Some good papers, some bad ones• More science, less prettiness (van Wijk ...)• Evaluations on the rise, but little in terms of

conclusions• No more trees, please!• Nice venue ;-)

Robert Kosara33/32

Next InfoVis

• October 28-29, 2002• Boston Park Plaza Hotel, Boston, MA

– Nice location– Bloody expensive– Pfister wants Biotech community there

• http://www.infovis.org/infovis2002