Collaborative visualization: definition, challenges, and research agenda

TL;DR: The purpose of this article is to help pinpoint the unique focus of collaborative visualization with its specific aspects, challenges, and requirements within the intersection of general computer-supported cooperative work and visualization research, and to draw attention to important future research questions to be addressed by the community.

Abstract: The conflux of two growing areas of technology - collaboration and visualization - into a new research direction, coLLaborative visualization, provides new research chaLLenges TechnoLogy now aLLows us to easily connect and collaborate with one another - in settings as diverse as over networked computers, across mobile devices, or using shared displays such as interactive waLLs and tabletop surfaces Digital information is now reguLarLy accessed by muLtipLe people in order to share information, to view it together, to analyze it, orto form decisions VisuaLizations are used to deal more effectively with Large amounts of information white interactive visuaLizations aLLow users to expLore the underlying data White researchers face many chaLLenges in coLLaboration and in visualization, the emergence of coLLaborative visualization poses additional chaLLenges, but it is also an exciting opportunity to reach new audiences and applications for visualization tooLs and techniques] to provide a definition, clear scope, and overview of the evolving field of coLLaborative visualization, (2) to help pinpoint the unique focus of coLLaborative visualization with its specific aspects, chaLLenges, and requirements within the intersection of generaL computer-supported cooperative work and visualization research, and (3) to draw attention to important future research questions to be addressed by the community We conclude by discussing a research agenda for future work on coLLaborative visualization and urge for a new generation of visuaLization toots that are designed with coLLaboration in mind from their very inception

Summary

1 Introduction

• Collaboration has been named one of the grand challenges for visualization and visual analytics,1 and for good reason: the problems that analysts face in the real world are becoming increasingly large and complex, not to mention uncertain, ill-defined, and broadly scoped.
• It is often no longer feasible for a single analyst to tackle the immense datasets that are now commonplace in the real world—realistic problems often require broad expertise, diverse perspectives, and a number of dedicated people to solve.
• Meanwhile, traditional visualization and visual analytics tools are typically designed for a single user interacting with a visualization application on a standard desktop computer.
• The emerging field of collaborative visualization is intrinsically interdisciplinary in nature, incorporating well-established research fields such as distributed computing, human-computer interaction (HCI), and, in particular, computer-supported cooperative work (or CSCW).
• Only then will the authors be able to identify the areas where they can best contribute and apply their knowledge and expertise.

2 Definition

• Previously, several definitions have been given to describe specific aspects of collaborative visualization.
• Similar to this restriction by type of collaborators, other definitions may have been too restrictive in terms of the applicable fields:.
• In order to more broadly describe the entire scope that collaborative visualization can encompass, the authors propose to define the term collaborative visualization as follows: Collaborative visualization is the shared use of computer-supported, (interactive,) visual representations of data by more than one person with the common goal of contribution to joint information processing activities.
• The theory of Group Cognition13 describes collaborative knowledge building for small groups by focusing on linguistic analysis, Distributed Cognition14 focuses on social aspects of cognition by analyzing the coordination between individuals and artifacts, and Communities of Practice15 describe learning within much larger social communities.
• The larger group involved in social interaction around data, for example, can simply view the information, actively interact with and explore it, or even join in creating new visualizations and share those and the underlying datasets with a larger community.

3 Research Background

• Research on collaboration started in the area of scientific visualization and, thus, many early tools focused on scientific datasets and techniques (e. g., using volume or flow analysis) and distributed synchronous collaboration in specific environments such as CAVEs or using head-mounted displays.
• This past focus is, for example, visible if one looks at the publications on collaborative visualization in the IEEE VisWeek conferences (Vis, InfoVis, and VAST).
• These three particular conferences were chosen as the top venues representing research interests of the larger visualization community but of course publications of collaborative visualization systems are also found elsewhere (ACM ITS, ACM CHI, and others).
• Out of 1583 papers published in the three IEEE VisWeek conferences—VIS since 1990, InfoVis since 1995, VAST since 2006—34 papers focused on collaborative visualization and only nine covered co-located collaboration.
• The following sections briefly outline two major research streams according to their main type of spatial collaboration scenario: distributed and co-located.

3.1 Distributed Visualization

• Within the area of distributed collaborative visualization , one research focus has been on architectures and synchronization mechanisms for allowing efficient synchronous remote work with large scientific datasets.e. g. 9,21–23.
• Much of this research is focused on applications in virtual reality (VR) over the web,e. g. 21 in GRID computing, e. g. 24,25 or for special hardware environments such as CAVEs.see 26 Grimstead et al.18 provide an excellent overview and taxonomy of 42 different distributed collaborative visualization approaches which describes and characterizes this stream of research in more detail.

3.2 Co-located Visualization

• Several other approaches have focused on the support of synchronous co-located collaboration with technology .
• Single-display technology often comes in the form of large interactive wallse.
• Additional synchronous inputs lead to new challenges.
• Two specific overview articles7,39 provide additional detail on the applicability of CSCW research on co-located collaboration to colocated collaborative information visualization.
• Not including changes made by SAGE after acceptance, also known as 43 4 Version 2.

4 Application Scenarios

• In the following subsections the authors provide five detailed real-world examples of scenarios in which collaborative visualization tools have been used.
• The authors outline the importance of dedicated visualization tools and techniques for specific work scenarios.

4.1 Collaborative Visualization on the Web: Many Eyes

• Many Eyes17 is a social data analysis website where people can upload, visualize, and discuss datasets using a set of pre-defined visual representations and a rich set of tools for annotation, feedback, and mashup.
• The stated goal of the website is to “democratize” visualization technology by exposing the technology to the broadest possible audience.
• Annotations are also linked to comments, but are used to highlight specific items within the state of a visualization (as opposed to the full state, as for a bookmark).
• Collaboration is also used for structuring the content on Many Eyes.
• This means that the registered users may collaborate not only directly on the Many Eyes website, but also have the ability to bring their visualizations, analysis and insights to their own online communities (e.g., social networks, forums, blogs) by embedding a visualization or a dataset in that context.

4.2 Collaborative Visualization for Scientific Research

• Many major science investigations such as high energy physics, computational chemistry, climate modeling, and astronomical studies are generating massive amounts of data that are stored in central or distributed storage repositories for sharing.
• The notion of “collaboratory”47 was introduced to support such largescale investigations.
• Scientists on this project may run the same simulation code with different parameter settings but do not only look at the output data they generated themselves but also those generated by others.
• When they examine the data, they can create visualizations and add notes summarizing their findings.
• By focusing on the dynamics of information exchange, Henline58 argues that the key challenges in creating a collaboratory may be social rather than technical.

4.3 Collaborative Visualization for Command and Control: Command Post of the Future

• CPOF is a computer system whose goal was to improve command and control using networked information visualization systems to double the speed and quality of command decisions.
• CPOF supports commanders with three key capabilities: (a) graphical views: 2D and 3D information visualizations, (b) information liquidity: drag-and-drop information analysis across different visualization products; and (c) topsight: visibility of evolving understanding among distributed subordinates and team members.
• Users share data but also can tailor their visualizations to capture the way they think.
• Figure 6 shows a user’s private space and a workspace that a number of users might share.

4.4 Collaborative Visualization for Environmental Planning

• Collaborative visualization in environmental planning has benefited from the confluence of two bodies of scientific literature that have rapidly emerged during the last two decades—Information Visualization and Collaborative Knowledge Construction.64.
• An example for a collaborative visualization tool for environmental planning is the synchronous and distributed collaborative geovisualization environment proposed by Brewer et al.62,63 and shown in Figure 7.
• The tool enables the exploration of climatic time series via interactions and animations, while offering support for collaborative sensemaking.
• One such example involves the use of the Solar Market Analysis and Research Tool that integrates disparate data related to deployment of solar power generation facilities in Arizona.
• Not including changes made by SAGE after acceptance, also known as 8 Version 2.

4.5 Collaborative Visualization for Mission Planning

• The MERBoard platform was built as a collaborative workspace for colocated scientists for NASA’s Mars Exploration Rovers (MER) mission.
• Collaborative work around SOLTree was typically conducted by small groups of three to 12 people, but the numbers decreased as the mission progressed.
• While collaborative authoring occurred, it also often happened that an individual would draft a plan alone and later gather other scientists around the display to discuss the plan and receive feedback.
• As such the application with its large-display setting in the workspace also served the purpose of a persistent information display providing community awareness.
• SOLTree is an interesting success story of a fairly simple visual planning tool which despite its simplicity in terms of visualization provided important support for a team of scientists in a highly collaborative, coordinated, and dynamic work environment.

5 Unique Focus of Collaborative Visualization

• As previously discussed, collaborative visualization lies at the intersection of two major research fields: traditional visualization and computer-supported cooperative work (CSCW).
• Clearly, both of these fields have long and rich histories, and the authors must be aware of both of these to make contributions to collaborative visualization.
• The position of collaborative visualization as a sub-area within two larger research fields brings a specific focus of its own.
• In the following the authors point out this unique focus from the standpoint of visualization research.
• The authors concentrate on the actual intersection between the fields (summarized in Table 1) to point out the unique challenges and requirements that require attention from researchers.

5.1 Users and Tasks

• From a core visualization standpoint, the most obvious focus of collaborative visualization is the addition of participants beyond the canonical single analyst that traditional visualization software is designed for.
• Having multiple participants is what transforms the analytical sensemaking12 process into a collaborative one and gives rise to all of the challenges discussed here.
• Similar to the broad range of CSCW research, collaborative visualization research has explored a range of group sizes starting from the basic paired-analysis scenarios39,72 to internet-sized audiences.73.
• The focus in collaborative visualization, however, is on a specific type of audience with either specific analysis questions and interests or even specific data-related background knowledge (“expert users”).
• This specific audience also carries specific tasks centered around visual representations and presentations of data ranging from specific work and/or domain-related data analysis questions to more open data exploration in museums or online.2 9.

5.2 Cognition and Results

• One of the main differentiating factors from the wider field of CSCW research is that the focus of collaborative visualization is often not the creation of a “product” (e. g., a photo layout or a text document) but an increased understanding or insight into a dataset, a consensus, or the ability to make informed decisions.
• The design of collaborative visualization systems poses challenges in addition to those encountered during the design of visualization systems that are intended to be used by a single person.
• A single person typically works with an information display through a process of viewing and possibly interacting with a visualization, forming a mental model by interpreting the representation, and ideally gaining an insight and forming a decision.
• Through social interaction (e. g., discussion and negotiation) they can also build on each others’ insights and potentially reach a common understanding of the dataset in order to make informed decisions as a group, derive common recommendations, or take next step actions together after the analysis.
• In their definition from section 2 the authors emphasize this contribution to joint information processing activities.

5.3 Interaction and Visual Representations

• One of the main challenges of visualization research is that data analysis is often a complex task: multistaged, poorly understood, and characterized by dynamic and conflicting information.
• In addition, within the process of data analysis, collaborative work can and does occur at different stages: information acquisition, representation and presentation, analysis and interpretation, sharing of analysis results, and making decisions and taking actions.
• Another research venue in collaborative visualization has been to find out how existing visual representations and interaction techniques need to be enriched and augmented to better support collaborative settings.
• These challenges are of both social and technical nature.
• Finally, whereas an objective within CSCW has long been exploiting novel computing hardware to support multiple inputs and output that facilitate collaboration, it is only recently that visualization research took the leap beyond the standard mouse-and-keyboard desktop computer.

5.4 Evaluation

• The success of visual data analysis is strongly connected to the mental model that a person forms about the data by viewing the visualization.
• The goal of using a collaborative information visualization system is typically to provide the group with an environment that enriches their data analysis activities beyond what they could come up with as separate individuals.
• Measuring group insight (or even individual insight as pointed out previously77) is difficult.

6 Collaborative Visualization Challenges and Research Agenda

• One of the main goals of research in collaborative visualization is to enable people to collaboratively use visual representations of data to gain additional understanding, knowledge, and insight into the data—different or more encompassing—than would have been possible had they explored the data individually.
• To learn more about how this goal can be reached, researchers have to address both the technical challenges of designing and implementing digital and physical environments that support collaborative data analysis, as well as the social aspects of group work.
• This section attempts to summarize a number of immediate goals and their own vision of the most urgent and promising directions and goals for collaborative visualization.

6.1 Address Dedicated Research Challenges

• In order for collaborative visualization systems to become used and adopted, it is important to solve interaction and representation challenges on all areas outlined in Section 5.
• Interactions with collaborative systems as well as the visual representations that are offered to a team are central to the abilities of each member to work with others, receive related information, and spontaneously react to emerging information and ideas from others.
• The authors need to learn more about how to design interactions and representations to specifically support collaborative reasoning and sensemaking.
• Only when collaborations are quick to set up, do not require considerable overhead to organize, and when the results of a collaboration can be quickly used to inform further work, will these setups be useful in practice.
• The authors therefore echo the recent call for more dedicated research on interaction with visualization systems1— particularly focusing on collaborative interactions and data exchanges.

6.2 Engage New Audiences

• As discussed early in the paper, a number of different analysis scenarios, needs, questions, goals, and challenges exist.
• To arrive at a more encompassing understanding and generalizable overview of collaborative visualization requirements and best practices, the authors need to ground their understanding in specific real-world examples.
• Within the next five years the authors therefore urge researchers to help in establishing connections to a wide 11 audience with collaboration needs, to study their collaborative needs and requirements, and publish reports on these investigations.

6.3 Standardize Collaboration Support

• At any of these stages, collaboration may be essential to ensure quality decisions, more encompassing solutions, or the integration of different viewpoints.
• Retrofitting visualization systems is possible75 but not always easy in retrospect, so providing collaborative support from the inception of a new tool will become increasingly important.
• ProtoVis78 and the InfoVis Toolkit75 are examples of toolkits where such an integration has begun.
• Within the next five years the authors hope to see this integration become a standard.

6.4 Expand to New Collaborative Spaces

• Collaboration can also occur outside and across the confines of the time-space matrix.
• In order to cover collaborative data analysis needs more broadly, the authors need to expand also to more research on hybrid collaboration scenarios.
• Kim et al.79 developed a toolkit called Hugin that supports collaborative analysis both on shared tabletop displays as well as across distance on remotely connected tabletop displays.
• Other hybrid scenarios (e. g., using the same space but used both synchronously and asynchronously) still need further research attention.
• Within the next five years the authors hope to see collaborative visualization research to have explored and addressed challenges of a number of different collaboration scenarios.

6.5 Develop Dedicated Evaluation Methods

• Developing dedicated methods for evaluating visualization systems has been an active topic of research in the last couple of years, and it is clear that assessing the value of visualization to people’s work processing, learning, or understanding of a topic is difficult.
• Similarly, the field of CSCW has been discussing how to evaluate groupware systems for a number of years and a variety of approaches have been proposed—yet never with a focus on assessing systems targeted towards data analysis with visualizations.
• The challenges have already been outlined above in Section 5.4.
• Within the next five years, the authors need to begin to develop new methods to assess the impact of collaborative tools and continue to refine and assess their value across different types of collaborative settings, data, tasks, or group sizes.

6.6 Integration and Adoption

• Distributed visualization systems17 have so far made huge progress towards integration and adoption of data analysis environments in collaborative settings.
• Several commercial systems have also begun to offer collaborative features, including Tableau Public, Spotfire Decision Site Posters,80 or MayaVis.81.
• Similarly, these audiences may be dealing with very different data characteristics.
• This stands in contrast to collaborative visualization for museum exhibits or shopping windows for broad audiences and data of an often lower information density.
• The authors hope that within the next ten years, research will have moved into practice in new venues and have shown to be successful, important, and enriching in a variety of situations.

6.7 Derive A Higher-Level Understanding

• One of the long-term goals of the community should be to derive a higher-level understanding of collaborative visualization challenges and requirements.
• But dedicated research in additional application areas will inevitably broaden their understanding and extend this initial set.the authors.
• As a community the authors need to encourage research on collaborative visualization, get students and young researchers interested in the topic, and continue 12 Version 2: Not including changes made by SAGE after acceptance.
• The authors need to learn about new and extended social as well as technical challenges to finally arrive at a higher-level understanding of characteristics of collaborative visualization which may span across different areas.

7 Conclusion

• The types of social exchanges around digital information can range from very casual online conversations with friends or family members about their social network, to discussions around museum exhibits, planned data explorations in research labs, to decision-making scenarios in conference rooms, or to internet-sized data explorations, discussions, and interpretations.
• Visualization of data will be central to the many collaborative interactions with digital information given its power in providing quick visual access to data and making information readily understandable.
• In order to enable and capitalize on this trend, it is important for visualization researchers to find out how the authors can make collaboration support a standard for data analysis environments.
• Numerous open research problems exist and in order for visualization to reach new audiences with their tools, solving these challenges will be essential.

Figures

Figure 5: An interface for visualizing accelerator simulation results shared by scientists in an internationally collaborative project. 46 Each point in the left half of the window represents a specific simulation run. By selecting a point, the user sees corresponding data in the form of images and animations as well as notes made by other users in the right half of the window

Figure 1: Collaborative visualization can occur in many scenarios delineated according to space and time.matrix adapted from 4,5

Table 1: Specific challenges to address in the research space intersecting collaborative work and visualization.

Figure 6: Command Post of the Future with a private and shared workspace (image c© General Dynamics Corporation).

Figure 7: The collaborative geovisualization tool by MacEachren and Brewer 62,63 showing two linked desktops that supported real-time collaboration with the application (figure courtesy of Alan MacEachren).

Figure 2: Distributed 18 and co-located 19 collaborative sensemaking. 20

Figure 8: Scientists using the SOLTree application at NASA (image c©Jay Trimble).

Figure 4: The ManyEyes web interface enables collaborative visualization of shared data. The main focus goes towards the visualization that users can create, customize and annotate with additional information and remarks. The comments section below captures the opinions of users about the data and the visualization, while at the same time allowing them to create bookmarks of the representation.

Figure 3: Papers published with a focus on collaborative visualization in three major visualization venues (IEEE Vis, InfoVis, and VAST). Shading and numbers above a bar indicate the number of papers on co-located collaboration for a venue per year.

Full text

Collaborative Visualization:
Definition, Challenges, and Research Agenda
Petra Isenberg
†
Niklas Elmqvist
∗
Jean Scholtz
¶
Daniel Cernea
‡
Kwan-Liu Ma
$
Hans Hagen
‡
†
INRIA
∗
Purdue University
¶
PNNL
‡
University of Kaiserslautern
$
University of California, Davis
Abstract
The conflux of two growing areas of technology—
collaboration and visualization—into a new research
direction, collaborative visualization, provides new re-
search challenges. Technology now allows us to eas-
ily connect and collaborate with one another—in set-
tings as diverse as over networked computers, across
mobile devices, or using shared displays such as inter-
active walls and tabletop surfaces. Digital information
is now regularly accessed by multiple people in order to
share information, to view it together, to analyze it, or
to form decisions. Visualizations are used to deal more
effectively with large amounts of information while in-
teractive visualizations allow users to explore the un-
derlying data. While researchers face many challenges
in collaboration and in visualization, the emergence of
collaborative visualization poses additional challenges
but is also an exciting opportunity to reach new audi-
ences and applications for visualization tools and tech-
niques.
The purpose of this article is (1) to provide a definition,
clear scope, and overview of the evolving field of col-
laborative visualization, (2) to help pinpoint the unique
focus of collaborative visualization with its specific as-
pects, challenges, and requirements within the intersec-
tion of general computer-supported cooperative work
(CSCW) and visualization research, and (3) to draw at-
tention to important future research questions to be ad-
dressed by the community. We conclude by discussing a
research agenda for future work on collaborative visu-
alization and urge for a new generation of visualization
tools that are designed with collaboration in mind from
their very inception.
Keywords: collaboration, visualization, computer-
supported cooperative work (CSCW), teamwork.
The final, definitive version of this article is pub-
lished in Information Visualization, 10(4):310–326, Oc-
tober/2011 by SAGE Publications Ltd, All rights reserved.
c
[The Author(s)]
1 Introduction
Collaboration has been named one of the grand chal-
lenges for visualization and visual analytics,
1
and for
good reason: the problems that analysts face in the real
world are becoming increasingly large and complex, not
to mention uncertain, ill-defined, and broadly scoped.
It is often no longer feasible for a single analyst to tackle
the immense datasets that are now commonplace in the
real world—realistic problems often require broad ex-
pertise, diverse perspectives, and a number of dedicated
people to solve. In addition, interaction with digital
information is increasingly becoming a social activity,
for example, on the social web or on large interactive
display technologies in public spaces
2
and visualization
research is only just beginning to expand its focus into
domains outside of the work environment.
3
Meanwhile, traditional visualization and visual analyt-
ics tools are typically designed for a single user interact-
ing with a visualization application on a standard desk-
top computer. Extending these tools to include support
for collaboration would clearly go a long way towards
increasing the scope and applicability of visualization
in the real world. However, the emerging field of col-
laborative visualization is intrinsically interdisciplinary
in nature, incorporating well-established research fields
such as distributed computing, human-computer inter-
action (HCI), and, in particular, computer-supported
cooperative work (or CSCW). As an outsider to these
fields, becoming familiar with their research in order
to start one’s own work on collaborative visualization
can be a daunting task; for example, CSCW research
spans 25 years and multiple conferences, journals, and
textbooks that have all advanced the field through the
years.
4
While collaborative visualization benefits from work in
other disciplines, there are many challenges, aspects,
and issues that are unique to the intersection of collab-
orative work and visualization. These are the places
where researchers have to play a significant role in ex-
panding the state of the art and help to shape where
and how visualizations will be used in the future.
In particular, CSCW research generally does not deal
with data analysis challenges coupled with interactive
1

The final, definitive version of this article is published in Information Visualization, 10(4):310–326,
October/2011 by SAGE Publications Ltd, All rights reserved.
c
[The Author(s)]
Time
Space
Co-located
Distributed
Synchronous
Asynchronous
Video conferences
Media spaces
Meeting room
Classroom
Museums
Lab space
Email-based data discussions
Web-based data analysis
Shift Work (e.g. hospitals)
Figure 1: Collaborative visualization can occur in many scenarios de-
lineated according to space and time.
matrix adapted from 4,5
visual data representations and much work remains to
be done to study collaborative data analysis, sensemak-
ing, and perception with and of visualizations in all
of the settings of the classic space-time matrix (Fig-
ure 1). Naturally, for this to be possible, visualization
researchers must first arm themselves with the prereq-
uisite knowledge, terminology, and culture that apply
from the CSCW field. Only then will we be able to iden-
tify the areas where we can best contribute and apply
our knowledge and expertise.
The purpose of this article is to help visualization re-
searchers with their investigations into collaborative vi-
sualizations. It is meant to be useful for those re-
searchers who may already have a background in col-
laborative visualization as well as those who are just
planning their first projects. The goals of this article are
(1) to provide a definition, clear scope, and overview
of the evolving field of collaborative visualization, (2)
to help pinpoint the unique focus of collaborative vi-
sualization with its specific aspects, challenges, and re-
quirements within the intersection of general computer-
supported cooperative work (CSCW) and visualization
research, and (3) to draw attention to important future
research questions to be addressed by the community.
We begin by discussing a broad definition of collabora-
tive visualization. We then study a set of representative
examples of areas where collaborative visualization—as
it fits our definition—has shown to be extremely ben-
eficial to data analysis: web-based collaborative visu-
alization, collaboration in scientific visualization, syn-
chronous collaborative visualization for dynamic anal-
ysis environments, and collaborative analysis for envi-
ronmental and mission planning. Drawing on this dis-
cussion, we propose a research agenda for future work
on collaborative visualization and to usher in a new
generation of information visualization tools that were
designed with collaboration in mind from their very in-
ception.
2 Definition
Previously, several definitions have been given to de-
scribe specific aspects of collaborative visualization.
None, however, have attempted to give an encompass-
ing definition of the entire scope of group work around
visual representations of data. In the following we dis-
cuss four previous definitions, note their limitations,
and finally provide our own definition for collaborative
visualization.
One of the earliest definitions emphasizes the goal of
collaborative visualization:
“Collaborative visualization enhances the tradi-
tional visualization by bringing together many
experts so that each can contribute toward the
common goal of the understanding of the object,
phenomenon, or data under investigation.”
6
While bringing experts together is an advantage in some
collaborative visualization scenarios, collaborators of-
ten do not need to be experts. Non-experts can join
in collaborative analyses and learn from others’ analy-
sis processes and viewpoints on a dataset.
7
Similar to
this restriction by type of collaborators, other defini-
tions may have been too restrictive in terms of the ap-
plicable fields:
“The term “collaborative visualization” refers to
a subset of CSCW applications in which control
over parameters or products of the scientific vi-
sualization process is shared.”
8
“Collaborative visualization [. . .] allows geo-
graphically separated users to access a shared
virtual environment to visualize and manipu-
late datasets for problem solving without physi-
cal travel.”
9
The first definition emphasizes collaboration with in-
teractive, manipulable visualizations for the scientific
visualization community. The restriction to only the
scientific visualization community is overly limiting as
the information visualization and visual analytics com-
munity similarly make use of collaborative systems to
analyze data. The second definition emphasizes dis-
tributed visualization in virtual environments. While
much of collaborative visualization research focused on
this area,
e. g. 10
groupware systems have a long tradi-
tion in both distributed as well as co-located spatial
domains. The limitation to virtual environments is an-
other unnecessary restriction. Collaborative visualiza-
tion also has had numerous applications outside of vir-
tual environments.
The restriction to only interactive visualizations in both
definitions may also be limiting and it is still being de-
2
Version 2: Not including changes made by SAGE after acceptance.

The final, definitive version of this article is published in Information Visualization, 10(4):310–326,
October/2011 by SAGE Publications Ltd, All rights reserved.
c
[The Author(s)]
bated whether interactivity should be a part of a general
definition of visualization.
e. g. 3
However, in this article
we only consider collaboration with interactive visual-
izations.
Recently, the term social data analysis has been coined
to describe the social interaction that is a central part of
collaborative visualization:
“[Social data analysis is] a version of ex-
ploratory data analysis that relies on social in-
teraction as source of inspiration and motiva-
tion.”
11
This term emphasizes the possibility of human interac-
tions such as discussions, negotiations, or arguments
around visualizations as the driving factors of data ex-
ploration. Yet, social interaction around data may oc-
cur in more scenarios than just exploratory data anal-
ysis. For example, targeted or confirmatory data anal-
ysis, teaching, learning, or decision-making scenarios
around visualizations may also frequently involve col-
laboration.
In order to more broadly describe the entire scope that
collaborative visualization can encompass, we propose
to define the term collaborative visualization as follows:
Collaborative visualization is the shared use of
computer-supported, (interactive,) visual repre-
sentations of data by more than one person with
the common goal of contribution to joint infor-
mation processing activities.
This definition is derived from a general definition for
visualization as the use of computer-supported, interac-
tive, visual representations of data to amplify cognition.
12
It has been augmented by emphasizing the shared use of
(interactive) visual representations—which could be in
the form of joint viewing, interacting with, discussing,
or interpreting the representation. Secondly, the term
“cognition” has been replaced with the term “informa-
tion processing.” This replacement acknowledges the
fact that different theories exist for how cognition ap-
plies when groups come together to jointly think and
reason. Each theory has different terminology, restric-
tions, and units of analysis. For example, the theory of
Group Cognition
13
describes collaborative knowledge
building for small groups by focusing on linguistic anal-
ysis, Distributed Cognition
14
focuses on social aspects
of cognition by analyzing the coordination between in-
dividuals and artifacts, and Communities of Practice
15
describe learning within much larger social communi-
ties. In order to avoid favoring any specific theory or
unit of analysis, we thus use information processing as a
general term to describe cognitive activities involved in
individual or collaborative processing of visual informa-
tion, such as reading, understanding, applying knowl-
edge, discussing, or interpreting.
Given this broad definition of collaborative visualiza-
tion, we can look at a number of different scenarios
in which it may occur. Using the space-time matrix,
4,5
we can broadly categorize collaborative scenarios ac-
cording to where they occur in space (distributed vs. co-
located) and in time (synchronous vs. asynchronous).
These distinctions for systems or tools are not strict—
systems can cross boundaries and could, for exam-
ple, be used both synchronously or asynchronously,e. g.,
rapid vs. long-term email exchanges.
5
Figure 1 shows
several scenarios in which collaborative visualization
can occur.
Another valuable categorization for collaborative vi-
sualization systems pertains to levels of engagement
teams have with a visualization system. The larger
group involved in social interaction around data, for
example, can simply view the information, actively in-
teract with and explore it, or even join in creating
new visualizations and share those and the underlying
datasets with a larger community.
16
Several digital sys-
tems have been designed to support collaborative visu-
alizations along these different levels of engagement, as
outlined below:
Viewing: Presentation systems such as PowerPoint or
simple videoconferencing tools can support a
group of people viewing static or animated visu-
alizations of data without being able to interact
with or annotate the information. Such scenarios
often occur, for example, in classrooms or meetings
where one presenter explains, teaches, or summa-
rizes information for the larger group. The goal of
the group may be to learn, discuss, interpret, or
form decisions from a pre-selected set of informa-
tion and visualizations.
Interacting/Exploring: When groups of people share the
same interactive visualization software, either in
co-located or distributed settings, they can choose
and select alternative views of the data for its ex-
ploration, analysis, discussion, and interpretation.
In distributed settings, findings can typically be
exchanged through chat, comments, e-mail, or a
video-/audio-link so that the changing views and
alternative representations of the data can be dis-
cussed and analyzed. This discussion can also oc-
cur face-to-face in co-located settings. The goal of
the group with this level of engagement is often
to cover and explore different and more aspects of
the data, consider alternative interpretations, and
discuss the data in a wider visual context.
Sharing/Creating: Through the emerging trend of user-
generated content sites for visualization (e. g., in
systems such as Many Eyes
17
), many people are
able to create, upload, and share new datasets
and visualizations. Often this type of sharing is
done within a greater community to raise aware-
ness about a certain issue.
Similar to the space-time matrix, levels of engagement
do not provide a clear-cut categorization of collabora-
3

The final, definitive version of this article is published in Information Visualization, 10(4):310–326,
October/2011 by SAGE Publications Ltd, All rights reserved.
c
[The Author(s)]
Figure 2: Distributed
18
and co-located
19
collaborative sensemak-
ing.
20
tive visualization systems. Digital systems may, for ex-
ample, be intended to mainly support collaborative in-
teraction and exploration of data but may also support
the sharing and creation of new visualizations or even
the download of new datasets to visualize. However,
both time and space dimensions as well as levels of en-
gagement can help to broadly scope a research focus
within collaborative visualization.
3 Research Background
Research on collaboration started in the area of scien-
tific visualization and, thus, many early tools focused on
scientific datasets and techniques (e. g., using volume
or flow analysis) and distributed synchronous collabo-
ration in specific environments such as CAVEs or using
head-mounted displays.
This past focus is, for example, visible if one looks at
the publications on collaborative visualization in the
IEEE VisWeek conferences (Vis, InfoVis, and VAST).
These three particular conferences were chosen as the
top venues representing research interests of the larger
visualization community but of course publications of
collaborative visualization systems are also found else-
where (ACM ITS, ACM CHI, and others).
Out of 1583 papers published in the three IEEE VisWeek
conferences—VIS since 1990, InfoVis since 1995, VAST
since 2006—34 papers focused on collaborative visual-
ization and only nine covered co-located collaboration.
Yet, in the past several years the support of collabora-
tive visualization has become increasingly important as
can be seen from the temporal trend in Figure 3.
The following sections briefly outline two major re-
search streams according to their main type of spatial
collaboration scenario: distributed and co-located.
3.1 Distributed Visualization
Within the area of distributed collaborative visualiza-
tion (left image in Figure 2), one research focus has
been on architectures and synchronization mechanisms
for allowing efficient synchronous remote work with
large scientific datasets.
e. g. 9,21–23
Much of this research
is focused on applications in virtual reality (VR) over
the web,
e. g. 21
in GRID computing,
e. g. 24,25
or for special
hardware environments such as CAVEs.
see 26
Grimstead
et al.
18
provide an excellent overview and taxonomy of
42 different distributed collaborative visualization ap-
proaches which describes and characterizes this stream
of research in more detail.
During the past several years, distributed web-based in-
formation visualization applications have emerged with
a focus on making information visualization accessible
to an internet-sized (mostly lay) audience.
e. g. 17,27
With
these systems, the research focus has shifted from the
more technical aspects of network latency, synchroniza-
tion, and view updates to more social, human-centered
questions such as how wide audiences can be engaged
to discuss and explore information, how laypeople can
effectively share data and visualizations online, or how
collaborative contributions can be effectively structured
and integrated into a shared visualization to ignite fur-
ther discussion and common ground formation.
7
3.2 Co-located Visualization
Several other approaches have focused on the support
of synchronous co-located collaboration with technol-
ogy (right image in Figure 2). These approaches can be
broadly categorized as those using single-display
28
or
multi-display technology.
Single-display technology often comes in the form of
large interactive walls
e. g. 29
or tabletop displays.
e. g. 30
Research in this area has, for example, described
mechanisms to support coordination of activities in
the workspace,
e. g. 31–33
awareness of group member’s
activities,
e. g. 34
or access to and transfer of items in
the workspace.
e. g. 35
With emerging display technolo-
gies such as multi-touch tabletop or wall displays, in-
dependent input for each group member becomes eas-
ier and cheaper to achieve without specific hardware
devices. However, additional synchronous inputs lead
to new challenges. Past research
36–38
has specifically
addressed how people can coordinate synchronous in-
put over visualization spaces. Two specific overview
articles
7,39
provide additional detail on the applicabil-
ity of CSCW research on co-located collaboration to co-
located collaborative information visualization.
Research on multi-display environments is concerned
with coordinating input and output from a number of
different display devices, such as large displays as well
as integrated mobile and wireless devices.
e. g. 40
Exam-
ples of past research endeavors include those for molec-
ular visualization across large displays and a tabletop
41
,
for geospatial visualization across a similar setup
42
, or
for a network setup which allows researchers to con-
nect and share their own visualizations from laptops on
large displays.
43
4
Version 2: Not including changes made by SAGE after acceptance.

The final, definitive version of this article is published in Information Visualization, 10(4):310–326,
October/2011 by SAGE Publications Ltd, All rights reserved.
c
[The Author(s)]
0
1
2
3
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
InfoVis
Vis
VAST
VIS
Since 1990
InfoVis
Since 1995
VAST
Since 2006
(1)
(1)
(1)
2009
2010
(2)
(2)(2)
Figure 3: Papers published with a focus on collaborative visualization in three major visualization venues (IEEE Vis, InfoVis, and VAST).
Shading and numbers above a bar indicate the number of papers on co-located collaboration for a venue per year.
Figure 4: The ManyEyes web interface enables collaborative visual-
ization of shared data. The main focus goes towards the
visualization that users can create, customize and annotate
with additional information and remarks. The comments
section below captures the opinions of users about the data
and the visualization, while at the same time allowing them
to create bookmarks of the representation.
4 Application Scenarios
In the following subsections we provide five detailed
real-world examples of scenarios in which collaborative
visualization tools have been used. With this section,
we outline the importance of dedicated visualization
tools and techniques for specific work scenarios.
4.1 Collaborative Visualization on the Web: Many
Eyes
Many Eyes
17
is a social data analysis website where
people can upload, visualize, and discuss datasets us-
ing a set of pre-defined visual representations and a rich
set of tools for annotation, feedback, and mashup. The
stated goal of the website is to “democratize” visual-
ization technology by exposing the technology to the
broadest possible audience. Because of its web-based
design, it is an example of an asynchronous, distributed
collaborative visualization tool: collaborators access the
website using their browsers through the Internet from
different places and at different times.
Many Eyes is a community-participation website, simi-
lar in multiple ways to other Web 2.0 sites like YouTube,
Flickr, Wikipedia, etc. However, unlike these sites, the
purpose of Many Eyes is to support several levels of
engagement including simple viewing of the data, in-
teracting with and exploring the data, as well as shar-
ing data (uploading) and creating new visualizations.
Accessing the website requires only a standard web
browser and a Java runtime environment.
The most important features of Many Eyes are its mech-
anisms for social sensemaking and collaboration. The
textual comment as the main communication mecha-
nism of the website can be added to any visualization
and dataset created or uploaded on the site just like
a user would comment on a blog post or in an online
discussion forum (see Figure 4). However, comments
and messages alone are not sufficient for establishing
common ground
44
necessary for efficient collaboration.
Many Eyes supports this process with two additional
features: bookmarks and annotations. A bookmark is
simply a snapshot of the full state of a visualization and
can optionally be stored together with a comment. An-
notations are also linked to comments, but are used to
highlight specific items within the state of a visualiza-
tion (as opposed to the full state, as for a bookmark).
The highlighting is simply done by selecting particular
items in a visualization when the comment is added.
Collaboration is also used for structuring the content on
Many Eyes. Because uploaded data on the site can come
from any area of interest (political, economical, tech-
nical, networking, etc), the site also supports group-
ing datasets and, implicitly, their visualizations into
topic centers. Similar to YouTube and other community-
participation websites, Many Eyes users have the possi-
bility to rate datasets and visualizations. This means
that the community also has a quantification role in
terms of correctness, as avoiding inaccurate conclusions
from faulty data or representations is highly desired. It
5

Frequently asked questions

Q1. What are the contributions in "Collaborative visualization:definition, challenges, and research agenda" ?

The conflux of two growing areas of technology— collaboration and visualization—into a new research direction, collaborative visualization, provides new research challenges. The purpose of this article is ( 1 ) to provide a definition, clear scope, and overview of the evolving field of collaborative visualization, ( 2 ) to help pinpoint the unique focus of collaborative visualization with its specific aspects, challenges, and requirements within the intersection of general computer-supported cooperative work ( CSCW ) and visualization research, and ( 3 ) to draw attention to important future research questions to be addressed by the community. The authors conclude by discussing a research agenda for future work on collaborative visualization and urge for a new generation of visualization tools that are designed with collaboration in mind from their very inception. 

Q2. What have the authors stated for future works in "Collaborative visualization:definition, challenges, and research agenda" ?

The Command Post of the Future60 ( CPOF ) was a Defense Advanced Research Project Agency ( DARPA ) project started in 1997. CPOF is able to dynamically incorporate new information which can be from data feeds or from user-entered data. This information can be viewed in different ways by the individuals consuming the information. CPOF is composed of modular components which can be scaled and tailored to fit different command environments. 

Q3. What is the role of collaborative visualization in environmental planning?

Computersupported collaborative visualization in environmental planning provides decision-makers the ability to (1) distill knowledge through mining large multidimensionaldata sets, (2) run models and simulations to explore the consequences of particular actions, (3) communicate results, scenarios, and opinions to other stakeholders, and (4) discuss, debate, and develop support for specific courses of action. 

Q4. What is the example of a collaborative visualization tool?

The tool enables the exploration of climatic time series via interactions and animations, while offering support for collaborative sensemaking. 

Q5. What is the example of a collaborative visualization system?

The SOLTree application, a part of the MERBoard70 platform, is an example of a successful co-located collaborative visualization system. 

Q6. What is the valuable categorization for collaborative visualization systems?

Another valuable categorization for collaborative visualization systems pertains to levels of engagement teams have with a visualization system. 

Q7. How have distributed visualization systems been incorporated into everyday environments?

Distributed visualization systems17 have so far made huge progress towards integration and adoption of data analysis environments in collaborative settings. 

Q8. What is the example of a collaborative visualization tool for environmental planning?

An example for a collaborative visualization tool for environmental planning is the synchronous and distributed collaborative geovisualization environment proposed by Brewer et al.62,63 and shown in Figure 7.