Report 1
The 2nd Workshop on Information Visualization was
a unique opportunity for some 113 participants from 8 government
agencies, 5 universities and some 46 companies and organizations
to exchange ideas and learn more about the leading-edge trends
in this domain. The program consisted of a keynote address
by Ben Shneiderman, a well-known leader in the field, followed
by seven invited, more specialized talks and a panel discussion.
Professor Ben Shneiderman, University of Maryland, HCIL
The Eyes Have It: User Interfaces for Information Visualization
Ben is the Founding Director (1983-2000) of the University
of Maryland, Human-Computer Interaction Laboratory. He set
the stage for the workshop with an enthusiastic and lively
description of the impressive progress achieved and of the
challenges lying ahead. Starting with an overview of the User
Interface Design Goals, he emphasized the importance of a
scientific approach centered on users and based on actual
measures. He then introduced Information Visualization as
one of the key opportunities that human visual perception
enabled. With a rich set of examples, he brushed a portrait
of several existing information visualization approaches and
their various applications. He explained how the visualization
task flow "Overview, zoom & filters, details-on-demand"
could be found in all of these. In conclusion, citing the
example of Spotfire, now used by 25 of the 25 top pharmaceutical
companies around the world, he described how today's information
visualization applications are becoming common and almost
indispensable tools in domains where maintaining finely tuned
knowledge is a determining success factor.
Martin Nilsson, Spotfire, Inc.
Analytical Applications for Networked Decision Making
Martin presented Spotfire. This visualization tool is based
on the Starfield approach initially developed at University
of Maryland, HCIL. Martin described how the ability to easily
import data makes it possible to build large unified databases
of unstructured data from multiple disparate sources. Using
networked role-based access, various teams inside the organization
can then explore these data to extract relevant knowledge
using interactive information visualization tools. The ability
to work in parallel to enrich the database and extract knowledge
is an essential asset as it greatly accelerates the process.
Ramana Rao, Inxight, Inc.
See and Go Wide Widgets for Navigation and Exploration
Ramana described how visualization tools can harvest the
power of a user's pre-attentive visual cognition abilities
and how they greatly surpassed classic "windows-icons-menus
and pointing" graphical user interfaces. He then demonstrated
with multiple examples how the "Star-Tree" (Hyperbolic Tree)
and the "Table Lens" visualization approaches enabled efficient,
quick and easy navigation and exploration of data sets following
a simple "See and Go" task flow.
Jeff Saffer, OmniViz, Inc.
A Cognitive Analytical Environment for Integrated Decision
Support with Multiple Data Types
Using the example of complex chemical data represented via
various approaches (Galaxy maps, Isoplots, etc), Jeff talked
about the exploration of data based on the creative visualization
of data sets composed of integrated information from different
sources. He stressed that no single view is sufficient to
really understand the data and that classic 2D graphs are
useless when thousands of items comprise the data set. Using
various views or coupled visualization approaches to explore
the data greatly eases the detection of trends, clusters or
outliers.
Eser Kandogan, IBM, San Jose
Visualizing Multi-dimensional data: Challenges and Opportunities
Eser compared Scatter Plots, Parallel Coordinates and Tables
Lens views to Star Coordinate representations to demonstrate
the superior flexibility of data massaging enabled by the
latter approach. Using the example of a Telecom Churn Analysis,
he focused his talk on the process of massaging data sets
by methods such as filtering and progressive systematic sub-setting
of the data to visualize trends, clusters or outliers.
Pierre Boulanger, NRC, IIT
How to Solve real World Problem Using Virtualized Reality
Pierre described how virtual models of real objects can be
studied using Virtual Reality Visualization to analyze their
behavior and monitor and control their interaction in their
environment. As an example, he also showed how Virtual Reality
Visualization can be used to represent a real object in order
to study it and optimize an aluminum die-casting process to
reproduce it.
Stephen Eick, Visual Insights, Inc.
Visualizing Website Activity
Stephen first described the problems of collecting data about
Website activity due to the varied nature of the information
and to the variety of sources. He demonstrated how this aggregated
information could be efficiently explored with various information
visualization approaches to discover trends and relations
visually and to measure activity according to metrics like
'Sessionization' or 'Repeat visitors'. Finally, he underlined
the importance of generating various sorts of reports going
from detailed analysis to executive overview.
Michael Kopko, SGI/FISC
Visual Data Mining in Finance
Michael outlined how visualization techniques could be used
to explore and comprehend complex interactions among risk
components in portfolios in a way that is meaningful for finance
experts. He stressed the complexity of the data by describing
the many dimensions: the sheer number of portfolios, the number
of trades per day, the tens of hundreds of positions, etc.
Once aggregated, this is displayed and explored using various
visualization techniques. The trends, correlations, outliers
and anomalies detected, plus the ability to check assumptions
and hypotheses, represent an invaluable intelligence that
has made strategic the visual mining of risk data in many
areas of finance.
Panel Discussion, Chaired by Martin Brooks, NRC, IIT
Martin's opening comments set the stage for clarifications
of topics that had been discussed by the panellists in their
presentations as well as for an interesting exchange relating
to other information visualization issues from the audience.
Program and other arrangements were cared for by the Organizing
Committee listed below. The assistance of the following sponsors,
is gratefully acknowledged:
- CODATA, Committee on Data for Science and Technology of
the International Council for Science;
- National Research Council of Canada, VP for Technology and Industry Support;
- Nortel Networks;
- Spotfire;
- Apple Canada;
- Allegra Solutions;
- CiriLab;
- Toth Information Systems;
- OCRI, Ottawa Centre for Research and Innovation;
- IEEE Ottawa Section.
Organizing Committee
Dr. Gordon Wood, National Research Council (Chair)
Dr. John Rodgers, Toth Information Systems (Chair, Program sub-committee)
Mr. Jean Paul Lauzon, AMEC Technologies, Sponsors Coordinator
Mr. Paul Amirault, National Research Council, Demonstrations Coordinator
Dr. André Vellino, Publicity Coordinator
Ms. Kathy Mahoney, OCRI, Local Arrangements
Ms. Terry D'Angelo, OCRI, Local Arrangement and Secretary
Ms. Marilyn Cheek, OCRI, Local Arrangements
1 Prepared
by Dr. L. Beaudoin, formerly Nortel Networks, and Dr. G. Wood, NRC.
|
