Interactive Posters Preview

1
Interactive Posters Preview

• Co-Chairs
• Alan Keahey Visintuit
• Tamara Munzner University of British Columbia

2
Towards an XML Toolkit for a Software Repository
Supporting Information Visualization Education
and Research

• Multiple visualizations of one data set can be
  generated simultaneously for comparison.
• Non-programmers with a basic understanding of
  the included algorithms can use a graphical user
  interface to visualize their data sets.
• Facilitates the design, comparison, and
  customization of IVs for learning the
  functionality of different layout algorithms.

Katy Borner and Jason Baumgartner Indiana
University - Bloomington School of Library
Information Science Available at
http//ella.slis.indiana.edu/katy/L697/code/
3
Design of the XML Toolkit

• The toolkit is structured through interfaces and
  a model that populates each layout algorithm from
  the XML file data.
• Enables the decoupling of the layout algorithm
  from the specific application layout
  instructions.
• Current direction The XML will be represented
  in a schema based on RDF, Dublin Core, and SVG.

4

• Documents anytime, anywhere
• Produced by people using mobile devices
• Laptops, PDAs, digital cameras
• Generated automatically in smart computing
  environments
• Meeting capture, Webcams, security
• Documents can be naturally associated with space
  time attributes
• Context
• Content
• How to use space time attributes to visualize
  documents for browsing and retrieval?

5

• Interactive Space-Time Maps
• Active regions on maps
• Semantic zoom
• Telescopic timeline
• SpaceTime Browser
• Retrieve documents in selected regions and time
  intervals
• Indicates uncertainty or relevance by fading
• Demo
• Room-aware documents
• Travel photos

6
Analytic Stimulus Response Animation
?Take an n-dimensional data set
? Each data entity is represented and visualized
as one Data Avatar
? Data Avatars move driven by a stimulus
response behavior model in a circular arena
Ralf Dörner Fraunhofer AGC
? Watch the Data Avatars move, interact with
them, manipulate their behavior model
Colin Ware University of New Hampshire
7

• Advantages of ASRA
• Encourage interaction and exploration of data
• more intuitive, more natural interpretation
• Human visual capabilities and domain knowledge
  exploited
• No pre-knowledge about data required
• Visualization of multi-dimensional data and
  fuzzy structures
• Distance metrics in feature space can be changed
  on on the fly with transition animation
  visualizing it

8
visualizing usenet with treemaps
INFOVIS2002
andrew t. fiore MIT MEDIA LAB
atf_at_media.mit.edu marc a. smith MICROSOFT
RESERCH masmith_at_microsoft.com
9

• Area indicates how many messages
• Color shows which areas have grown, declined

10
A zooming model for geographic maps Daniele
Frigioni Laura Tarantino University of LAquila

• We want to support
• LOD approach
• detail-on-demand
• many foci
• focus context
• fish-eye views

11
METHODS AND RESULTS

• conceptual model of data based on posets
• kernel of elementary zooming primitives
• traversability of the map space
• advanced multiple zooming primitive
• topological invariants bridge the gap between
  spatial data and visual interface
• incremental algorithms to generate maps

12
Visualizing Egocentric Social Networks by Douglas
Gordin, dgordin_at_us.ibm.com
13
Visualizing Egocentric Social Networks by Douglas
Gordin, dgordin_at_us.ibm.com
14
Visualizing E-Business and Web Services Using
Pixel Bar Charts Ming C. Hao, Daniel Keim,
Umeshwar Dayal, Fabio Casati, Akhil Sahai, Vijay
Machiraju Hewlett Packard Research Laboratories,
Palo Alto, CA
Average is high, but large of trans are fast
day 4 has slowest response time

• Bar charts are widely used, but only show
  aggregated data. Finding the valuable information
  hidden in the data is essential to the success of
  e-business

fast response time ( 0.013 sec)
Shows data distribution in detail

• Pixel bar charts integrate basic bar charts and
  X-Y diagrams with the following attributes
• dividing (for between-bar, Dx , Dy )
• ordering (for within-bar, Ox , Oy)
• coloring (for pixel coloring, C)

Presently with University of Konstanz,
keim_at_informatik.uni-konstanz.de
15
Pixel Bar Chart Applications 63,544 business
process instances
1. Business Process Distribution Analysis
3. Service Level Agreement Analysis
(A) Analysis of SLA Status (failed, passed)
duration gt47days
status
violation
0.6
0.5
(A) Dx ProcessName, Oy duration, C of days
04
0.3
02
0.1
drilldown from above failed bar
(B) Dx hour, Oy duration, C of days
sla
(B) Drill down on failed SLAs SLA 1000 has the
highest violation levels (color dark brown and
burgundy)
2. Web Service Correlation Analysis
(B) Dx date, Oy CPU utilization, C value
of I/O rate
(A) Dx date, Oy CPU utilization, C value
of CPU utilization
16
Visual Queries for Finding Patterns in Time
Series Data
Timeboxes rectangular queries for time series
data
17
TimeSearcher
Now available for academic and Non-commercial
use.
http//www.cs.umd.edu/hcil/timesearcher
18
Visual Data Mining With The Interactive
Dendogram.
Peter Imrich Klaus Mueller Ray Mugno
Dan Imre Alla Zelenyuk Wei Zhu

• Supports
• Hierarchical clustering of high-dimensional data
• Multi-resolution viewing
• Interactive user-control over clustering
• Data reduction via pre-processing

19
Graphical User Interface
Global visual search
Drilling into a cluster / node
Group selection (time slice, etc.)
20
Addressing Scale and Context in Source Code
Visualization
Versioning
Features
Design patterns
Profile data
Data flow
Syntactic structure
Use cases
Control flow
Documentation
21
Addressing Scale and Context in Source Code
Visualization

• 3D Source-Quake Arena
• Provides smooth tradeoff between scale and detail
• Affords code browsing
• Projection model based on semantic fisheyes
• Addresses multivariate data
• Provides reusable goal-based views

• Semantic zooming for source code
• Increase semantic content of distant code
• Increase contextual information

S. Jackson, K. Ma, P. Devanbu
22
Visual Information Retrieval with Bubble
WorldTim Jacobs and Chris Van BerendonckAir
Force Institute of Technology

• Interactive visual techniques for exploring large
  unstructured data collections
• Visual interface for search engine
• Based on cognitive principles and usage patterns
  from the information retrieval domain
• Combines multiple techniques for
• Visualizing and exploring retrieved results
• Graphically refining queries

23
Experiment with a live demo during the poster
session
24
Dancer
Streaming Graphics in Real Time Andrew Norton and
Leland Wilkinson SPSS Inc. Northwestern University
25
Dancer
Exponential smoothing of (live-feed) stock series
for Oracle and SPSS
26
Ringtuples Model
State Diagram
Dt1
Dt2
Dt3
ON
OFF
t
Time Sequence
M time value for bins m number of bins
Binning of intervals ti floorDti/M
Vectorial Value
Vk v0, v1, v2, , v2m-3, v2m-2, v2m-1
Past States
Present States
27
Ringtuples Visualization
Total Time in the ON State is mapped to the area
of a circle
3D
2D
Rings
Side by Side Comparison of Ringtuple Icons
28
(No Transcript)
29
(No Transcript)
30
(No Transcript)
31
(No Transcript)
32
(No Transcript)
33
(No Transcript)
34
Visualizing Protein-Protein Interactions on a
Genomic Scale

• Interactions between proteins in the cell form
  the basis for all meaningful work in the cell.
• Now that biology is in the post-genome era, the
  focus has shifted to determining what all the
  proteins do.
• First draft of the human genome project indicates
  between 35,000 and 50,000 genes.
• High throughput methods are creating databases of
  protein-protein interactions.
• Automated visualization methods are not
• equipped to handle genome-sized data sets.

35
Visualizing Protein-Protein Interactions on a
Genomic Scale

• ProInAct a better way to visualize interactions

36
MONDRIAN - Interactive Data Vis in JAVA

• MONDRIAN implements Interactive Statistical
  Graphics
• Barcharts
• Mosaic Plots
• Histograms
• Scatterplots / Binned Plots
• Parallel Coordinates / Boxplots
• Maps
• All Plots are fully linked and support a unified
  interface
• Two data sources are supported
• ASCII files
• JDBC connections

Martin Theus, Augsburg University
mondrian_at_theusRus.de
37
MONDRIAN - Interactive Data Vis in JAVA

• MONDRIAN implements the novel Selection Sequences
  to navigate through complex data sets
• All plots can cope with very large data sets
  (gt100.000) either by applying
• Logical Zooming or
• a-blending
• MONDRIAN is written in 100 pure JAVA and runs on
  
• Mac OS X and other UNIXs
• Windows (except NT)
• Come, see the Poster !!

Martin Theus, Augsburg University
mondrian_at_theusRus.de
38
Illuminating Data Analysis
Ulrich FahrnerRobert Schmied

• What Explore captivating questions
• HowData set from last Soccer World Cup
• Why Demonstrate interactive features

39
Concepts
Interactive video to explain data analysis

• About the use of interactive statistical software

From the vantage point of a soccer expert
40
IMMERSIVE INFORMATION MODELING USING PARTICLES

• exploratory information visualization
• for 3D, immersive, stereoscopic, virtual reality,
  cave environment
• remote database objects represented by particles
• information particle infoticle
• tools, interface, interaction
• filters user queries ? user-defined spatial
  regions of equal data value
• forces user interests ? attract infoticles with
  same data value

41
IMMERSIVE INFORMATION MODELING USING PARTICLES

• modeling
• explore data by creative modeling in 3D space
• features
• data large amount, different dimensions, real
  time streams possible
• interaction intuitive, direct, real-time
• interface as changeable 3D objects placed in
  virtual scene
• goal
• dynamic properties (speed, direction, ) form ?
  informational values

42
XMDV-Multidimensional Visualization

• Solution to High Dimensional Datasets
• Group Similar Dimensions into Dimension Hierarchy
• Navigate Dimension Hierarchy using InterRing
• Form Lower Dimensional Spaces using Dimension
  Clusters
• Convey Dimension Cluster Information using
  Dissimilarity Display

43
XMDV-Multidimensional Visualization

• Solution to Large Scale Datasets
• Group Similar Records into Data
  Hierarchy
• Navigate Data Hierarchy using Structure-Based
  Brushing
• Represent Data Clusters using Mean-Band
  Method
• Provide Database Backend Support using MinMax
  Tree, Caching, Prefetching

Two solutions work together to visualize high
dimensional, large scale datasets!
44
Metavisualization of Dynamic Queries

• A metavisualization is a visualization of another
  visualization.
• Our focus is metavisualization of interactive
  structure, particularly linking in dynamic query
  visualizations.
• Improvise is an editor and browser for dynamic
  query visualization and metavisualization.
• Views are directly linked using Live Properties.
• Views are indirectly linked using Dynamic Query
  Expressions.
• Our poster describes how we have used Improvise
  to metavisualize DEVise visualizations.

45
Metavisualization of Dynamic Queries
46
Variable Density Scroll AreasFor Rapid Image
Retrieval

• Problem Design sketch to explore browsing
  large visual datasets
• Approach Display all data, all the time
• Use RSVP to display current data
• Maintain context for the user
• Allow user to control browsing activity
• Solution Use horizontal dimension to control
  density of large information spaces

47
Variable Density Scroll AreasFor Rapid Image
Retrieval
Discrete scroll area
Continuous scroll area
48
Web Site Visualization Using a Hierachical
Rectangle Packing TechniqueYumi Yamaguchi,
Takayuki Itoh, Yuko Ikehata, Yasumasa Kajinaga
Red web pages accessed in a hour
49
Flowchart of our website visualization tool
access statistics
access log file
sitemap
hierarchical website data
50
Online Social Spaces

• Lack social context information
• Who are the participants?
• What happened recently?
• What are the hot discussions?
• Content-centric view not ideal for social
  browsing
• Social browsing can enable participation
  social networking

Jun Zhang, School of Information, University of
Michigan, Alison Lee, IBM T.J. Watson
Research Center
51
eTree

• Visualization of social context of conversation
  spaces
• Information about people, activities, and social
  interaction
• A life-like tree ecosystem metaphor
• Browse and query interface
• Mechanisms for awareness, chat, and context
  comprehension
• Integrated into real online social space ? IBM
  Interns ? Portkey
• Demonstrate metaphor, operation, visualization
  for Portkey

Jun Zhang, School of Information, University of
Michigan, Alison Lee, IBM T.J. Watson
Research Center