Framework for spatial visual design of abstract information

Abstract

Spatially-organized information can be accessed and operated on rapidly and effortlessly, especially when a spatial arrangement reveals the conceptual organization of information. Therefore, spatial perception plays an important role for cognitive processing when interacting with abstract information. The process of spatial information visualization is shaped by various factors including interactive, perceptual, navigational as well as organizational and metaphorical aspects and as such requires an interdisciplinary approach. Consequently, bringing the knowledge from different disciplines requires the development of a framework which can host and classify the interdisciplinary features important in designing effective spatial visualizations. In this paper we present a framework which manifests a holistic approach in designing spatial visualization of abstract information.

Full text

Page 1

1
Framework for Spatial Visual Design of Abstract Information.


Malgorzata Bugajska
University of Zurich, Department of Informatics, Switzerland
bugajska@ifi.unizh.ch


Abstract
Spatially-organized information can be accessed
and operated on rapidly and effortlessly, especially
when a spatial arrangement reveals the conceptual
organization of information. Therefore, spatial
perception plays an important role for cognitive
processing when interacting with abstract information.
The process of spatial information visualization is
shaped by various factors including interactive,
perceptual, navigational as well as organizational and
metaphorical aspects and as such requires an
interdisciplinary approach. Consequently, bringing the
knowledge from different disciplines requires the
development of a framework which can host and
classify the interdisciplinary features important in
designing effective spatial visualizations. In this paper
we present a framework which manifests a holistic
approach in designing spatial visualization of abstract
information.

Keywords: spatial visualization, framework,
information visualization


1. Introduction
Spatial perception plays an important role for cognitive
processing when interacting with abstract information,
since spatially organized information can be accessed
and operated on rapidly and effortlessly, especially
when a spatial arrangement reveals the conceptual
organization of information. Therefore, an important
benefit that improves quality of processing abstract
data is the incorporation and use of spatial schemas
while designing visual representation. It becomes even
more crucial today when, due to the popularity of
handheld devices, the size of the display is constantly
decreasing whilst the amount of information which is
expected to be displayed on the digital device is
increasing.
In this paper we describe a framework which
embraces features important for spatial interactive
design of abstract information. This framework has
already been successfully used to organize spatial
design guidelines for information visualization [see 6].
2. Importance of spatial design for
information visualization
Space is a crucial dimension of our everyday life. In
space, we perceive and recognize objects and relations
between them. In space, we manipulate these objects
and we can move around to observe them. Depicting
space has been used for a long time to convey concrete
ideas. However, only recently it is being used to
convey abstract ideas [38]. Tversky [38] notices that
spatial schemas, by linking together elements, provide
an organization which improves memory and can
sometimes be a more powerful organizer of memory
than time. Additionally, spatial manipulation is a
largely subconscious activity that imposes very little
cognitive load, hence offering very powerful
functionality [21].
Abstract data is lacking inherent spatial mappings,
and additionally, the relationship between the data
value and the data view is multi-faceted [9, 10] As a
result, it is challenging to create a spatial set up for this
type of data since it requires applying interactively
different views of the same data set or applying an
operation of data spatial-filtering to compare different
data sets. Consequently, effective spatial representation
of data requires understanding the phenomena
governing the perception of space.
3 Holistic approach to spatial visual
information design
Spatial visualization depends on many aspects
relating to data attributes and organization, design
method, and available display technology. There is a
need for organizing components required for guidance
in spatial, interactive visualization and for investigating
relations between them. The nature of such relations
depends on a visualization goal envisioned by the
designer and the cognitive tasks to be fulfilled by the
user. Additionally, it is important to be aware of
various levels that are involved in shaping the quality
of spatial information visualization output. Some of
these levels include perceptual, interaction and

Page 2

2
navigation levels as well as organizational or
metaphoric levels. Until now, these aspects were
investigated separately by various researches [7, 19,
26, 28, 29, 35, 36, 37]. Many frameworks focusing on
different portions of design space have been proposed
[3, 8, 9, 11, 22, 25, 27, 33, 39]. Many classifications
have been created to help designers develop their
works [10, 20, 23, 29, 36, 43]. Additionally, principles
and rules for different aspects of information
visualization design have been developed [2, 3, 4, 8,
17, 35]. Understanding the spatial visualization process
in the digital domain requires an awareness of the
holistic nature of the act of space perception.
Furthermore, the process of reusing and sharing design
expertise should be structured and open to facilitate
sharing competencies among different design domains.
Therefore, we propose a framework which manifests a
holistic approach to designing spatial visualizations of
abstract information. Additionally, an important part of
the framework that classifies spatial visual cues is
based on analysis of spatial visual properties for digital
domain (see [6] for more details and examples).
4 Framework for spatial visualization
design and its elements
The framework presented here integrates various
aspects of spatial visualization. It encompasses
elements of visualization pipeline as described in [12]
by incorporating aspects of mapping data to geometry,
assigning visual properties to geometry and integrating
the user’s visualization tasks together with the goals of
the visualization designer.
representation of the framework (Appendix 1) we mark
the contributions of various researchers with “patches”.
Properties and tree-structures of properties without a
patch represent our contribution to the framework. We
take a top-down approach to describe the main
elements of the framework.
During the process of assigning properties to
visualization objects we distinguish five elements,
which define our “Digital Visualization Space”. Three
elements create the core of our classification:
•
Object – a graphic element or geometry which
is used to represent concepts in our world
•
Context – a graphic space which is used to
represent relations between elements [38]
•
Order – definition of a choice of spatial
arrangement of objects in the graphic space
In the graphical
Figure 1. Five main groups of spatial visualization
factors build framework for the spatial visualization
of abstract information: User Task Group, Designer
Goals Group, Context Group, Order Group, and
Object Group. Additionally, relationships between
Context, Order, and Object groups are visualized as
lines connecting spatial design factors on different
levels.
Designer Goals and User Tasks are two further
classification elements which have an important
influence on spatial visualization effectiveness. The
influence of these components is represented as a circle
embracing the Context, Order and Object elements in
the graphical representation of this classification. For
defining The User Tasks element, we adopted the
cluster of tasks method formulated by Shneiderman
[37] who defined seven items of “Information-Seeking
Mantra” while working with the visualization
environment. Please note that the visual examples of
spatial framework elements can be found in [6].
4.1 Elements of classification model -
Object Group
Within Object Group, we embrace three levels
describing information visualization objects. We
describe a visualization object in terms of its role
within a visualization scene using taxonomy of
graphical Simple Marks, Compound Marks, and
Negative Space. Furthermore, we acknowledge the
classification of the visualization object as a data
object, process object and referential object. Finally,
we classify visual properties of a visualization artefact
which enriches the spatial visualization of abstract
information.

Page 3

3
Visualization object: graphical marks and negative
space
Marks first introduced by Bertin [3] refer to graphical
elements visible on a display medium; they describe
the most primitive blocks which encode information:
points, lines and areas (volume mark was added later
by [8]). Senay [35] further extended the ordering of
graphical elements by adding another group of marks
called compound marks. They define Compound
Marks as “collections of simple marks that form a
single perceptual unit”. On this level of classification
we additionally introduce Negative Space which, in our
opinion, is an important element of information
visualization artefacts. Marcus [28] observes that
“empty” or negative space is needed to provide
emphasis for visual elements within the display space.
Also, Arnheim [1] points out that the relationship
between figures can only be understood if the spaces
between them are designed as carefully as the figures
themselves.
Data, Process, and Referential Objects
In this framework, we classify objects in terms of roles
they play within a visualization artefact. We
distinguish between three types of objects: Data
Object, Process Object and Referential Object. By
Data Object, we refer to any object that visually
represents different types of data. Process Object refers
to the type of object that supports interaction processes
between a human and a machine on the visual level.
We refer to Referential Object, which is based on the
concept of Referential Component [35], as any type of
visual object facilitating the proper interpretation of
spatial qualities in a graphical scene but not encoding
data items directly. The characteristics of a Data Object
are determined by the type of data being visualized.
We use the seven data types introduced by
Shneiderman [37].
The Process Object group clusters Icon, Label,
Filter, and Menu elements. We understand Icon as an
object that graphically represents an item recognizable
or learnable by the user. Icons can be used for
communicating certain functions or processes within
visualization artefacts. We refer to Label as an object
attached to another object (data or process-related)
describing this object,
representation. Filter represents types of object used
during the process of exploring visualized content for
modifying the spatial and graphical parameters of
objects (e.g. changing spatial configuration of objects
according to the new rule established by the user).
Finally, Menu refers to a wide range of processes
collected in one set of functions available for use when
working with visualization artefact. Within Object
often using textual
group we classify visual properties that influence
spatial design in visualization of abstract information.
For more detailed description of these properties (with
examples) please refer to [6].
Spatial Properties
We have grouped selected properties for spatial
information visualization
organized groups. A group clusters properties of
visualization artefact that share the same aim in
supporting spatial visualization on a flat display or a
group collects properties bearing similar perceptual
characteristics to evoke a spatial illusion in a
visualization artefact.
1. Position Group clusters properties that assemble
techniques which visually define the position of the
objects within a spatial environment. Orientation
property points at the position of an object with respect
to the observer’s viewpoint for reading the spatial
qualities of this object. Dropped Line property
describes the use of supporting elements to clarify the
geometrical position of visualization elements within a
spatial scene. Visual Frame communicates the
influence display borders have on reading spatial
relationships within the visualization artefact. Gravity
is an example of a force that can visually influence
characteristics of the behaviour of visualization objects
within the spatial scene. Elevation describes elements
acting as reference for objects which are subjected to
gravitational force within a visualization artefact.
Finally, Distance Association illustrates the technique
of defining the distance between objects based on the
observer’s knowledge of the size of these objects.
2. The Sequence group clusters properties
supporting qualities of sequential occurrences in space
and time. Rhythm describes a time-bounding property
helpful in prolonging the user’s attention to the visual
presentation. Disintegration illustrates an art of
composing elements on different planes in a spatial
organization. Finally, Level of Detail conveys a
technique that governs the detail’s order of appearance
within a visualization artefact. Detailed descriptions of
these properties follow.
3. The Layering Group clusters properties
describing the effects of relationships between objects
positioned on the same visual surface. Occlusion
describes spatial quality derived from the superposition
of objects. Figure/Background
importance of the relationship between objects and the
background for spatial visualization. Finally, Shape
Contour illustrates spatial effects that can be achieved
by manipulating shapes of contours on the surface.
4. The Symbolic Form Group describes techniques
that control spatial organization. Pyramidal Space
in nine thematically
underlines the

Page 4

4
represents
environment using rules of linear perspective. Negative
Space illustrates issues related to the impact of an
empty space on the spatial environment. Finally,
Distortion presents an approach for manipulating the
shape of objects with the goal to create a spatial scene.
5. Properties clustered in Kinetics Group explain
techniques for conveying time-based events within
computer-generated environments. Temporal Sequence
describes the process of integrating time into the
presentation of a phenomenon. Point of View describes
a technique that involves the user in a time-based
exploration of the visualization environment.
6. The Gradient Group depicts techniques that
evoke the perception of space by differentiating the
size or density of objects positioned near the observer
from those positioned far away. Relative Size Gradient
uses properties of familiar size for spatially positioning
elements. Textual Gradient illustrates the use of
elements building the texture in a spatial environment.
Finally, Relative Density describes the use of density of
objects as an aid for supporting the spatial quality of a
visualization scene.
7. Properties in Brightness Group stress the role of
light in modelling the spatial environment. Value
describes the relationship between lightness and
darkness of a colour. Shadow property explains the
importance of shadow in defining spatial environment.
Lighting and Colour properties illustrate the role of
brightness in a spatial composition. The property of
Transparency describes the effect of seeing through the
objects involved in a spatial scene.
8. The Focus Group explores the effects of
controlling the cognitive and vision-related focus of the
user when viewing a visualization artefact. Depth of
Focus describes how the illusion of depth can be
modelled using blurred and sharp objects. The Law of
Visibility presents rules that have an impact on a spatial
visualization. Atmospheric Perspective illustrates the
modelling of the spatial environment using different
types of visual gradation.
9. The User Group collects properties of User
Control and Continuity of Illusion which define the
importance of the user’s engagement in the spatial
visualization environment. User Control focuses on the
possibilities of the user for exploring the spatial
organization of the visualization artefact. Continuity of
Illusion refers to the spatial quality of a visualization
artefact.
These nine groups are further classified into clusters
which organize groups of properties according to the
role they play within a spatial visualization artefact.
Based on [3] vocabulary, we classify spatial properties
into Positional Cluster, Temporal Cluster, and Retinal
Cluster. Positional Cluster groups spatial properties,
a technique for depicting spatial which describe character of the visualization scene
depending on the spatial position of the object and
relationships between objects with respect to the user’s
viewpoint. Temporal Cluster describes techniques for
depicting space and time-based events within
visualization environments. Finally, Retinal Cluster
assembles properties which visually define spatial
character of the object - its shape, the effect of light on
it, or density of its texture.
Connection and Closure group are located on the level
of Positional, Temporal, and Retinal Clusters in our
classification and are described in more detail in [6].
During the design process of a computer-generated
spatial visualization of abstract information, it is
important to consider the impact of user action and
technology on the visualization artefact. Therefore, we
include in our classification additional clusters of
properties: User Defined Cluster, Input Device Cluster,
and Display Cluster.
Input Device Cluster differentiates between two types
of devices: a natural and a synthetic one. We refer to
the natural device as a device which facilitates the
transfer of instructions or information into the
computer for processing or storage, using natural
human capabilities for communicating through voice,
touch, gesture, body movement, eye contact, etc.
Synthetic devices require from the user a special type
knowledge of how to operate them (e.g. to enter
instruction into the computer with help of mouse,
keyboard, joystick, etc.)
Display Cluster includes items influencing the use of
other visual spatial properties by the choice of display
parameters such as size, resolution, and the use of
colour. The higher the resolution, the more complex
visual information can be presented on the display.
Compare, the spatial character of Linear Perspective
which can be diminished if the resolution of the display
is low, since the diagonal lines may appear jagged and
do not to convey the effect of objects disappearing in
the space behind the display.

There are three strong categories emerging from the
organization of spatial properties proposed until now.
Connection, Closure and Position clusters group
properties underlining the value of spatial relationships
between objects. Temporal and Retinal clusters
concentrate on describing the relationship between the
spatial scene and its meaning. Finally, User Defined,
Input Device, and Display clusters include properties
illustrating the impact a presentation device and user
involvement have in a spatial presentation. To
complete our classification of visual spatial properties,
we use visual semiotics. Using the approach of [16]
and [28] we distinguish three categories: Visual
Syntax, Visual Semantics and Visual Pragmatics.

Page 5

5
Visual Syntax refers to the quality of arrangements of
signs used for spatial communication. It therefore
includes Connection and Closure as well as Position
Clusters.
Visual Semantics deals with relationships between
signs and what they refer to [5]. In our classification,
spatial properties that stress relationships between the
type of technique chosen for creating a spatial scene
and the communication purpose of this visualization
scene belong to this category. We include Retinal and
Temporal clusters into this category. Finally, Visual
Pragmatics deals with the relation between the signs
and their users [5]. Similarly to [28], we extend this
definition to include technical conditions influencing
the process of communicating the signs. For the
purpose of our classification we include User Defined,
Input Device, and Display clusters into this category.
These clusters group properties that define the
limitations of visual spatial communication (e.g. size
and resolution of the display) as well as direct the
influence the user has on the spatial visualization scene
(see Appendix 1).
4.2 Elements of the framework - Context
Group
Perception is not an entirely stimulus-driven process.
Our perception is influenced to some extent by
“cognitive constrains: higher level goals, plans, and
expectations” [32]. Here we propose to include factors
that influence context of spatial visualization. We refer
to them as contextual factors. Until now, we have only
explored their visual features in spatial information
visualization environments.
Contextual factors describe components affecting
user’s spatial exploration of information on the
following levels: user, community, and environment
levels. “User level” represents the attributes a cognitive
model user is using when interacting with a particular
visualization. Cognitive attributes describe the task and
role of the user in the visualization environment.
“Community Level” - defines characteristics of a
digitally-based social space formed in a multi-user
visualization setting.
encompasses visualization components or methods that
define and characterize an environment dedicated for
user exploration in a perceptible way. The following
elements belong to this group: ‘spatial container’,
‘orientation’, ‘interaction’, and ‘expression’. Spatial
container embraces a structure that visually identifies
the user’s exploration space. Orientation represents a
set of characteristics defining the act of navigating
through the information space. Interaction describes
the group of techniques, which allow the user to
“Environment level”
influence visualization. Finally, expression specifies
the metaphorical concept used for spatial visualization.
User level - Mental Model
Mental Model is an important component of user
interface. After [42] we refer to Mental Model as a
basis for “understanding the system, for controlling its
action and predicting its future behaviour”. Mental
Model represents the organization of data, functions,
activities, and roles that users inhabit within computer-
based environments of work or play. We distinguish
two factors of the Mental Model group which have
been already introduced by [13]: Pattern of Presence
and Pattern of Association. Pattern of Presence is a
mental map (or model) of the user’s presence within a
visualization environment. Pattern of Association is
understood as a conceptual map created by the user
representing the community formed within the user’s
spatial system which would otherwise have no visible
manifestation in the physical world.
Community level - Social Space
Social Space is a group of factors defining social
aspects of a spatial multi-user environment. This group
includes: Digital Portrait, Digital Conversation, Digital
Crowd, and Social Networks. After [13] we refer to
Digital Portrait as a representation of the user within a
spatial multi-user environment. Digital Portrait has a
visual form which demonstrates the user’s presence as
a member of social space (through the actions he
provokes and takes) within the environment. As
already noted by [14], Digital Portrait depicts “a
culture as well as an individual which tells far more
about its subject than just what he or she looked like”.
Digital Conversation in newsgroups, chatrooms, wikis
and mailing list forms is referred to by [14] as the
foundation of Social Space in online environments.
Digital Conversation describes space and time-
dependent conversation taking place between users of
an online environment (synchronous and asynchronous
conversations). Digital Crowd, an expression coined
by [30] describes visualization of users simultaneously
visiting spatially-defined environments of online
documents or websites. Minar [30] lists three important
requirements for visualizing a digital crowd: a map of
the digital environment (e.g. documents visited by
users) for providing a spatial structure for a digital
crowd, a representation of individual users to show
visitors of the site, and an animated demonstration of
crowd dynamics. Crowd dynamics describe methods
for distinguishing between popular and less popular
groups of websites by mapping users accessing or
leaving these websites. Finally, Social Networks
(netvis.org) are patterns of relations or connections