|
Information
Visualization (IV):
Notes about the 8th IV ’04 International Conference, London,
England
Anna Ursyn (bio)
University of Northern Colorado
Ebad
Banissi
London South Bank University
Abstract
Information
and knowledge visualization is an area of focus within a visualization
community and involves a wide spectrum of other disciplines
(such as humanities, psychology, sociology, and business) due
to the progress in developing tools that incorporate 2D, 3D,
interactive, and virtual reality. Several international conferences
and symposia add every year a fresh view and thought to this
field.
This
review tells about the International Conference on Information
Visualization that is held annually in London, England. The
leading theme chosen for the 8th Information Visualization Conference
(IV 2004) was “To bridge the gap between visual imaging
and decision making.” Themes reported at this Conference
related to the progress in visualization, information visualization,
visual data mining, multimedia, web graphics, graph theory application,
augmented and virtual reality, semantic web visualization, digital
art, HCI, among many other areas such as information visualization
in geology, medicine, industry and education.
A
small number of themes have been selected from the IV ’04
Conference Proceedings (IV 2004, Proceedings of the 8th International
Conference on Information Visualization, 2004, IEEE, Los Alamitos,
CA), mainly those that have been focused on connections with
digital art and aesthetics, semantic approach to visualization,
and also some applications of information visualization.
Introduction
Visualization
means converting data into visual forms, thus showing objects
that are easy to recognize and connected through well-defined
relations. Designing a spatial mapping of numerical data sets
can be done through creating metaphors and expressing meaning
through images. Auditory and other sensory representations are
also of concern. Visualization tools reinforce cognition, as they
allow the opportunity to acquire knowledge about concepts, events,
or processes, the ability to locate information, and explore the
structure. Computers convert data into picture forms: simple charts
and graphs or virtual reality environments happening in real time,
with human interaction possible through various input/output devices,
which let us fly around the organized data, comprehend and make
decisions.
Information
visualization helps to explore abstract data, select, transform
and then represent them as static or interactive (mostly visual)
representations. It helps understand abstract data and their complexity,
manage, communicate and navigate the data, for instance, on the
web. Thus the progress in information visualization can be seen
as a part of the developments in a broader field of visualization.
Information visualization techniques provide 2D and 3D interaction
metaphors that are used for web navigation and communication.
This practice builds on theory in information design, computer
graphics and graphic design, art, semiotics, human-computer interaction,
cognitive psychology, and statistical graphics.
Data
mining. Data and processes to be visualized are often chaotic,
non-linear, going on various levels and in global dimension. Big
and complicated sets of data, for example, statistics about weather
or some government issues are easier to grasp in graphic form
than when expressed verbally or in numbers because graphics serve
as explanatory tool for these data sets. Data sets, such as stock
market or satellite telecommunication are too huge to be gathered,
stored, processed, and distributed. Visual data mining tools help
to organize information by integrating concepts, applying visualization
metaphors and methods, information and scientific data visualization,
diagrammatic reasoning, and building 3D collaborative virtual
environments for information visualization or advanced Virtual
Reality (VR) 3D visualization.
Visualization
techniques reveal the patterns and take up metaphors that could
be shared with the users. As Ben Shneiderman (IV 2004, p. 5) stated
in his keynote lecture, designs of the tools for multi-dimensional,
hierarchical or other visualization are often accepted as understandable
but sometimes perceived as confusing. For this reason, insights
for designers and usability tests may provide cognitive hypotheses
and practical guidelines for developers.
1.
Aesthetics and Design of the Data Visualization Sets
The
application of aesthetics to the field of visualization supports
a balance between form and function that is essential in the human
interaction with the computer. This part of the review relates
to the aesthetics-related issues that involve some principles
of designing the data set visualization. It gives a brief outline
of essays and practical applications: a discourse about representing
the complexity of information structures, the use of knowledge
visualizations by architects, information on the coding of knowledge
in architectural structures, an attempt to express the beauty
of molecular models, creation of an image stream installation
as a philosophical cave of illusions, and finally, the use of
ambient information visualization, that sits on the periphery
of a user’s attention, in service of aesthetical presentation
of the data.
1.
1. Aesthetics for visualization design
Looking
for perfection in visualization design involves advancing aesthetics
and finding conceptual inspiration to bridge the gap in visualization
discourse between science, art, technology and design. As stated
by Greg Judelman (IV 2004, pp. 245-250), http://isnm.de/~gjudelma/,
an interactive visualization is an interface to a complex data
space. Between the data layer and the graphic layer lies software
code, to map data into the visual domain. Visualizations should
be designed to transmit the maximum amount of information with
the minimum cognitive exertion of the user. Strategies for visual
communication can be find in other areas, for example, guidelines
for graphic design, 3D architectural visualization, animation
and interactivity in media arts, or cartographic maps using spatial
metaphor translated to informational one. The challenge is in
developing modes of abstraction to represent complexity of information
structures. The use of metaphors is often based on cartography
or natural environments. Metaphors are structurally complex but
universally familiar: users know the symbolic relationship between
visual or interaction elements.
Judelman
provides pointers to resources that can provide aesthetic and
conceptual inspiration for visualization design and generate ideas
for metaphors and interaction models. His examples come from algorithmic
art, architecture (e.g., centrifuge.org/marcos), cross-disciplinary
computational techniques, e.g., involving evolutionary systems
– artificial life, (such as alife.org, generative.net),
cellular automata, and nature. The author discusses algorithmic
or generative art as a sub-specialty of interaction design. He
points to several programming platforms in which practitioners
experiment with creative modes of visual interaction, including
Macromedia Flash and Director for accessing web-based applications,
and the non-proprietary Processing (processing.org) for coding
Java applets visualizations. For
3D graphics, real-time works are produced in VRML or game
engine environments, while a ray tracing software (povray.org)
serves for non-interactive video work. Audiovisual processing
environments suggested by Judelman include Jitter
and Keyworx. Virtual architectures
provide spatialization of information in virtual environments
where the users navigate VRML interactive environments to browse
information (for
example an Infotube). Visualizations that use natural elements
as a metaphor include:
StarryNight
and Spiral,
ThemeRiver,
and
DataCloud.
The
challenges facing information visualization researchers, as well
as those working in the science, arts, or design areas, involve
finding innovative graphic or interactive techniques to represent
the complexity of information structures.
1.
2. The use of visualizations by architects to amplify cognition
and transfer knowledge
Remo
A. Burkhard (IV 2004, pp. 519-524) clarifies the difference between
the research on information visualization and knowledge visualization.
Information visualization is a study on the use of computer-supported,
interactive visual representations of abstract data to amplify
cognition. It explores abstract data to create new insights. Knowledge
visualization examines the use of visualizations for the transfer
of knowledge between at least two persons or a group of persons.
It examines information overload, misinterpretation and misuse
of the information. Architects use five main visualization types:
sketch, diagram, image, object (physical model), and interactive
visualization. The author discusses the knowledge visualization
framework that is important for an effective transfer of knowledge.
Three
perspectives are described, that are used to amplify cognition,
to process visual representations, and to transfer knowledge:
(a) The
knowledge type perspective that identifies the 5 types of knowledge:
declarative (know-what), procedural (know-how), experimental
(know-why), orientational (know-where), and individual (know-who).
(b) The
recipient type perspective that identifies target group (an
individual, team, organization, network of persons with some
known cognitive background)
(c) The
visualization type perspective (sketch, diagram, image, object,
interactive visualization).
The
author presents an interactive knowledge visualization cube, implemented
in VRML that integrates the three perspectives for choosing visualization
method. Slices, then a single cube can be selected, with the use
of a link that loads detailed information.
1.
3. Visualization in architecture
Strategies
for visualization design can be find in 3D physical environments
created throughout the history of architecture in various cultures
in the form of built structures with encoded information. Peter
Ferschin and Andreas Gramelhofer (IV 2004, pp.181-186) discussed
how some design principles for spatial coding of information can
be extracted from architectural design and transferred to the
field of information visualization. Before the times of computing
human beings used images to communicate, sometimes on the walls,
so architecture was included into communication media. A term
‘information architecture’ that relates now to the
structure of a website, can describe instructive function of architectural
design.
Information
on various fields of knowledge has been coded in architectural
structures.
Examples provided by the authors include:
–
Proportion, using metaphors of human body, is presented in cathedrals
and Egyptian temples. Examples can be found in the works of
Le Corbusier, Di Giorgio Martini, and Schwaller de Lubicz.
–
A metaphor of music can be found in some architectural structures.
Representation of Bela Bartok music in the Steven Holl’s
Stretto House (1991) is an example of the relationship between
music and architecture.
–
Sacred knowledge of ten stages of enlightenment symbolized on
the ten terraces of Borobudur (Java), 760 a.d.
–
Cosmological order has been reflected in Balinese architecture
–
Astronomical and physical science of the time was visualized
in Jaipur in 1730 a.d.
–
A memory theatre, about 1550 a.d., provided a prototype of information
architecture as a metaphor of different categories of knowledge.
–
Architecture served as the presentation media using buildings
as surfaces for displaying information: video screens, engraved
texts, lighting, and advertisement tables.
–
Architecture has been often augmented with flat screens and
electronic tickers for floating information, the combination
of physical and virtual content. An example of augmented architecture
is the Asymptote’s
design for the Advanced Command Center (ACC) of the New
York Stock
1.
4. Scaled molecular models and sculptures
The
rules for creating models of natural objects make possible to
generate artistic forms of great complexity and order. Edgar F.
Meyer (IV 2004, pp. 229-234) searched for the minute beauties
of nature and aesthetic expression of scientific concepts. He
presented molecular models sculpted in wood, applying computer
graphics to define input parameters to a computer program SCULPT
which generates a code for a computer numerically controlled milling
machine. Scaled models of molecules have been carved in wood,
such as birds-eye maple
or walnut.
The author comments, “Computer graphics image lasts 1/30-1/60
sec. (the display refresh rate) whereas a wooden model will last
at least 500 years with reasonable care.”
1.
5. Plato Cave: an image stream installation within an office setting
Suzanne
M. Marchese and Francis T. Marchese (IV 2004, pp. 954-958) provided
persons working in windowless cubicles with an active cityscape.
They installed a video – a view of the surrounding neighborhood.
It was streamed from a web camera through a network and projected
on a hallway wall at Pace University, NY. The authors discuss
their piece as a philosophical cave of illusions – a simulacrum
of a street – in terms of the concept of Plato’ cave.
As for the changes induced by the differences in lighting, the
authors recognize artistic influences of paintings by Claude Monet
and Camille Pissarro.
1.
6. Ambient information visualization
Ambient
information visualization has been described as the use of computer-supported,
interactive, visual representations of abstract data to amplify
cognition, and also as the use of aesthetically pleasing displays
of information that sit on the periphery of a user’s attention.
Xiaobin Shen and Peter Eades (IV 2004, pp. 285-289) introduce
ambient information visualization in service of aesthetical presentation
of the data. Previous approaches to this kind of visualization
include the use of digital family portraits to display family
health issues, virtual paintings as InfoCanvas, or Mondrian-like
installations about bus traffic conditions. The authors provide
an interactive visual decoration that shows stock price and volume
data, and inform the user of changes in this data. They display
tree images to represent stock price changes and volume data.
They also applied triptychs using morphing technique, from one
tree to another. In one frame, they present number of stocks,
and each stock volume, prize, framed as an artwork, where trunks
show trade volume and leaves show the trade prize.
2.
Approaches to Managing the Data
2.
1. Event-based approach to visualization
Users
are usually interested in various aspects of the same data set.
Christian Tominski and Heidrun Schumann (IV 2004, pp. 101-107)
apply the brushing of data records in their event-based approach
for visualizing data sets. As an example of an event, a threshold
event of more than 300 cases of influenza is described. The authors
applied an axes-based visualization framework. VisAxes –
three interactive axes serve for data exploration: a scroll axis
to select a value range, a focus+context axis to emphasize certain
points of interest, and a hierarchical axis to structure value
ranges.
2.
2. Protection of the data visualization images
Shiguo
Lian, Jinsheng Sun, and Zhiquan Wang (IV 2004, pp. 217-220) apply
JPEG-based encryption in service of image protection. To protect
image data, authors present an image encryption algorithm combined
with JPEG encoding that is of high security, low cost, and can
be decrypted even if its compression ratio has been changed. JPEG
encoding (loss mode) realizes bit-rate conversion by changing
quantization steps.
2.
3. Accessing repositories for learning, education or training
Learning
objects, entities that serve for learning, education or training,
are often stored in Learning Objects Repositories (LORs) such
as Ariadne Knowledge Pool
System (KPS), Merlot,
and EdNa (http://www.edna.edu.au/edna/page1.html). Information
visualization techniques may enable better search for learning
objects than Boolean combinations for search criteria. Joris Klerkx,
Erik Duval, and Michael Meire (IV 2004, pp. 465-470) discuss and
compare three visualization techniques they have applied to the
Ariadne: Squarified Cushion Tree-Map, Hyperbolic tree, and Venn
diagrams with a Grokker program.
2.
4. Augmented reality for presenting information
Fotis
Liarokapis, Martin White, and Paul Lister (IV 2004, pp. 761-767)
propose a high-level augmented reality (AR) interface toolkit
that allows the combination of audiovisual information with a
real world environment, to enhance human perception and interaction
with the real word. Some examples of already developed AR systems
are: the Magic Book - uses a real book to transfer users from
reality to virtuality; an AR tangible interface is a tabletop
system designed for virtual interior design (users can interact
with virtual furniture and manipulate objects); the Studierstube
Personal Interaction Panel (PIP) - a blank physical board on which
virtual controls are drawn.
The
Mixed Reality Interface Toolkit can be applied for interior design,
medical, education, learning and training systems. The AR Interface
Toolkit can build new applications, having an audio-visual augmentation,
realistic object augmentation (e.g., with augmented shadows),
image augmentation, textual annotations, and audio augmentation.
Museum application for augmented representation of virtual objects
is an example of possible uses of the system. Future augmentations
include virtual touch, with haptic devices, such as virtual gloves.
3.
Semantic Approach to Visualization
3.1.
Approaches to define ‘semantics’
According
to Dirk Burmeister, Stephan Grimm, and Joerg Haist (IV 2004, pp.
302-307), ‘Semantics’ can be defined as an (intensional)
approach to define internal relations in a network of concepts.
This approach lacks an ontological anchor in realism. People can
communicate on a network of concepts and misunderstand because
they rely on this network without the need of hinting at real
objects. To rely on real objects as semantic carriers of concepts,
a second, extensional definition is needed that refers to speech-external
sections of reality. The symbols (words) are labels for data.
The third definition of semantic relies on the use of concepts
and the role of the speaker. The extensional view is seen as a
denotative use of concepts, and intensional view as connotative
use of a concept. The separation between user profile and ontology
is not necessary any more. The objective here is to get better
user involvement instead of dividing semantic into an ontological
part and a functional part processed by not semantically based
subcomponent of the system. With making the processes of the system
semantic, it is possible to extend semantics to wider fields of
semantic carriers as only words. The task is to map semantic roles
with their situation-dependent usage to proper metaphors. For
example, the sign language can extract such metaphors because
gestures encode both intensional and situation-dependent visualizations
of content.
Burmeister,
Grimm and Haist discuss how techniques in modeling semantic content
can be transferred from industrial application to the requirements
of design for all. WIDE, information retrieval software, uses
Semantic Web technologies to interpret and process queries and
produce shared information sources for engineers and designers.
A three-level approach is used to enrich given data to access
them in a semantic way. WIDE is the system that adopts the user’s
terminology, so no user has to learn the system terminology. The
search is supported by metadata, semantically annotated content
and ontologies, to make user queries understood by the underlying
information sources. The result of query is semantically enriched
by the system and translated to the user’s terminology.
Four
components of WIDE architecture are:
- The user
interface that presents results and their semantic dependencies
mainly using visualization techniques
- The Meta
level which performs semantic processing like interchange of
synonyms, interpretation of semantic processing and keyword
matching
- The agency,
which maps the business logic of the whole application using
user-group dependent agents to adapt to profiles of predefined
user groups
- The content
level, with sources for information to integrate: internal and
external databases, intra- or Internet with http-access.
3.2.
Perception based visual syntax and semantics
Some
theorists argue there is innate predisposition for understanding
linguistic structures (Chomsky, Pinker). Theories of perceptions
hold there are deeply seated mechanisms for decomposing visual
objects and analyzing them into both component parts and the structural
interrelationships of those parts. The perceptual theory of Marr
(1982) contains visual primitives and a visual syntax to understand
3D objects. The rules of visual structures are similar to verbal
language rules (Jackendoff, 1987) and there are cognitive ‘correspondence
rules’ between visual meaning and linguistic structure.
Perception based visual syntax and semantics should also support
accelerated learning. Pourang Irani and Colin Ware (IV 2004, pp.
308-314) propose the theory that diagrams, called geons, that
activate the mechanisms for structural object perception should
be similarly easy to learn, remember and analyze. The authors
developed diagramming principles based on the theory of structural
objects perception and recognition proposed by Marr. Geon diagrams
that activate the mechanisms for structural object perception
should be easy to learn, remember, and analyze. The diagrams are
characterized by nodes (represented as boxes and circles) representing
entities, objects, or units, and by interconnected links (represented
as lines) showing relationships of various kinds. First, objects
are decomposed. The next step is the geons’ structural description
that relates to geons, their attributes, and their relations with
adjacent geons.
The
UML (Unified Modeling Language) is used for modeling software
systems and is capable of modeling through its diagrams. Diagrams
help communicate. Although the semantics used are general and
complete, the choice of graphic notation is arbitrary and may
be less effective as teaching aids.
An
empirical study evaluates the learnability of geon diagram semantics
in comparison with the UML (Unified Modeling Language). Both ‘novices’
and ‘experts’ found the geon diagram syntax easier
to apply in a diagram-to-textual description-matching task than
the equivalent UMC syntax. Average error rates of matching diagrams
to problem descriptions were much lower as related to the Geon
diagrams than the UML diagrams. The use of geon notation could
be extended to other software engineering semantics.
4.
Conclusions
Topics
selected for this review do not reflect the whole field of visualization,
information visualization, and visual data mining. They do not
cover the broad spectrum of the specialized sessions held at the
8th International Conference on Information Visualization in London,
for example, computational and graphical tools and techniques,
issues related to human computer interaction, or a great number
of papers focused on medical visualization. This cross-section
of selected themes reflects a one-year input to the areas of interest
and disciplines. Further progress is reflected in the proceedings
of the successive conferences.
References
Burkhard,
R. A. (2004). Learning from architects: The difference between
knowledge visualization and information visualization. Proceedings
of 8th Int. Conf on IV, IEEE, Los Alamitos, 519-524.
Burmeister,
D., Grimm, S., & Haist, J. (2004). A semantic approach for
user depending information visualization. Proceedings of 8th Int.
Conf on IV, IEEE, Los Alamitos, 302-307.
Ferschin,
P., Gramelhofer, A. (2004). Architecture as information space.
Proceedings of 8th Int. Conf on IV, IEEE, Los Alamitos, 181-186.
Irani,
P. & Ware, C. (2004). The effect of a perceptual syntax on
the learnability of novel concepts. Proceedings of 8th Int. Conf
on IV, IEEE, Los Alamitos, 308-314.
Judelman,
G. (2004). Aesthetics and inspiration for visualization design:
Bridging the gap between art and science. Proceedings of 8th Int.
Conf on IV, IEEE, Los Alamitos, 245-250. http://isnm.de/~gjudelma/
Klerkx,
J., Duval, E. & Meire, M. (2004). Using information visualization
for accessing learning objects repositories. Proceedings of 8th
Int. Conf on IV, IEEE, Los Alamitos, 465-470.
Lian,
S., Sun, J., & Wang, Z. (2004). A novel image encryption scheme
based on Jpeg encoding. Proceedings of 8th Int. Conf on IV, IEEE,
Los Alamitos, 217-220.
Liarokapis,
F., White, M., & Lister, P. (2004). Augmented reality interface
toolkit. Proceedings of 8th Int. Conf on IV, IEEE, Los Alamitos,
761-767.
Marchese,
S. M. & Marchese, F. T. (2004). Plato Cave: an image stream
installation within an office setting. Proceedings of 8th Int.
Conf on IV, IEEE, Los Alamitos, 954-958.
Meyer,
E. F. (2004). Scaled molecular models and sculptures. Proceedings
of 8th Int. Conf on IV, IEEE, Los Alamitos, 229-234.
Shen,
X., & Eades, P. (2004). Using money tree to represent financial
data. Proceedings of 8th Int. Conf on IV, IEEE, Los Alamitos,
285-289.
Shneiderman,
B. (2004). Keynote lecture. Tools for multidimensional, hierarchical,
and other kinds of IV. Proceedings of 8th Int. Conf on IV, IEEE,
Los Alamitos, 5.
Tominski,
C., & Schumann, H. (2004). Event-based approach to visualization.
Proceedings of 8th Int. Conf on IV, IEEE, Los Alamitos, 101-107.
|
|
|