A Service Oriented Architecture to Visually Explore Large Data Collections
Romain Vuillemot, B
eatrice Rumpler and Jean-Marie Pinon
LIRIS, INSA Lyon, Universit
e de Lyon, F-69621, Lyon, France
Visualization On-Demand (VizOD), Information Visualization (InfoVis), Service Oriented Architecture
Facing the relentless information volume increase, users are not only lost in information overload, but also
among the various ways to depict it. In this paper, we tackle this issue by providing end-users access to up-
to-date visualizations techniques, using remote services coupled to their local interactive environment. The
outline of a Visualization On-Demand (VizOD) architecture is introduced, packaging information visualization
processes into services reachable over a network. Our goal is to provide end-users flexible and personalized
visual overviews of large datasets. We implemented a prototype partially validating our architecture, and
discuss preliminary results of our experiments and give future work perspectives.
In this paper we are interested in making visualiza-
tion techniques broadly available to end-users accord-
ing to his data, needs and task to achieve (see Fig-
ure 1). Availability is meant in terms of time-to-
product, skills required and span of choice to offer
users the right visualization technique, coupled into
their local interactive environment (software and in-
teractive devices), at the right time.
Figure 1: General user’s data, needs and tasks.
Painted with broad strokes, the major technical is-
sue with visualization techniques is the data format
heterogeneity and the lack of reliability: there is a
huge gap between a proof of concept issued from re-
search works and a out of the shelve tool. Advanced
contributions exist, but scattered in so many different
application fields. For instance, visual data mining
tools are very prolific in biology (Adai et al., 2004)
with stunning results facing real life problems, espe-
cially dealing with data masses. New heuristics of
finding patterns in huge structures are available for
a specific tasks, but those inspiring data depiction re-
main domain specific. At the end of the day, end-users
cannot benefit from most of the scientific tools even
if they look inspiring and potentially useful. And fi-
nally, concerning interactive environments, the desk-
top metaphor is only still massively used because
of universal availability: innovation cannot make its
As companies massively digitalize data for pro-
ductivity, ubiquity and quality management, users
would like to follow the behavior of massive and
complex data evolution, coming from many sources.
Whereas there exists tools to perform dedicated
analysis of the data, as far as we know there is no
way to get a visual overview carrying insights that
will trigger a more complex investigation with a dedi-
cated interface, regardless of the data or task (see Fig-
ure 2). In other words, there are no generic visual-
ization setups dealing with both accurate and global
information. Our goal is to uncover phenomena un-
Vuillemot R., Rumpler B. and Pinon J. (2008).
ANATOMY OF A VISUALIZATION ON-DEMAND SERVER - A Service Oriented Architecture to Visually Explore Large Data Collections.
In Proceedings of the Tenth International Conference on Enterprise Information Systems - HCI, pages 86-93
DOI: 10.5220/0001715000860093
seen at the first sight, which will guide the user to a
specific data subset or data projection.
Existing solutions are either basics visualization
techniques such as sorted lists or unorganized items,
never surpassing ordinary tables or spreadsheets.
Thus, optimized access is only available when users
change context by switching execution environment.
But if no solutions are available then a whole prod-
uct life cycle is started, inducing cost and time waste.
This process also runs the risk of decreasing user’s at-
tention and productivity. New approaches have to be
considered to keep users in the same interactive en-
vironment, with the same interactive devices he mas-
ters, just with a switch of data being analyzed.
Figure 2: Generic overview (gray area) of data structures
leads to more specific analysis.
Actors needs are identified as follow:
End-User Needs. There is a need for abstract
data handling solution to get an easy and quick
visual insight of the data, with appealing visual
metaphors. Such a process has to get rid of key-
board or not requiring any symbol being entered
into the system (in case user cannot formalize his
needs). Users have different background richness
and cultures, so individual accessibility character-
istic must be taken into account, such as visual
Designer Needs. They are experts in the field
of translating user needs into software or assem-
bly of software coupled with interactive devices.
Whereas design are stored in a guideline format
(Shneiderman and Plaisant, 2004), they lack for-
malization and evaluation. Re-use of existing li-
braries and environments becomes crucial with an
increasing systems complexity and heterogeneity.
The life cycle of products has to be etended as
Managers Needs. Responsibilities involve ad-
vanced monitoring tools with high reliability.
Trends are a way to anticipate the future and can
raise by means of complex interactions analysis or
long term data integration.
This paper is organized as follow. Section 2 fo-
cuses on the two major scientific fields being bridged,
that are Information Visualization and Service Ori-
ented Architecture. Section 3 focuses on the archi-
tecture outlines. Section 4 describes a prototype that
has been developed. Section 5 discusses results and
perspective. Section 6 concludes.
Our approach deals with Informations Visualization
(InfoVis) techniques, which technical lacks are to be
covered by web services packaging included in a Ser-
vice Oriented Architecture (SOA). A synthetic com-
parison is given on Table 1, and similar attempts are
also listed and analyzed.
2.1 Information Visualization
Conceptually, the fundamental goal of InfoVis is to
find the right visualization at the right moment. A
complete visualization process results in outputs such
as maps to discover interrelationships between data.
Relationships can be either internal or external, help-
ing to understand complex static or dynamic dataset
growing over time or actions. Human capabilities are
thus enhanced, but limited cognitive memory must
be taken into account in order not to overload users
with information. The next step is for codes or user’s
knowledge to be integrated to reduce information di-
mension. Limited display space must also be con-
sidered, that can be solved by coordinated multiple
views, either from a static perspective which consists
of two or more distinct views supporting the investi-
gation of a single conceptual entity (Michelle et al.,
2000). Or either dynamically by synchronizing dis-
tinct views over time and/or user actions (Shneider-
man, 1996).
Technically, the underlying problem is that visu-
alization is hard coded to data and task to achieve.
Today, reusing a technique for other data means start-
ing another configuration/implementation cycle ac-
cording to informal design guidelines. These recom-
mendations lack formalization and are given in pat-
tern format which has to be understood by experts,
inducing extra costs. Another limit is that visualiza-
tions are local to user’s application and dependent of
his computing power. And because contributions are
scattered in so many different application fields, there
ANATOMY OF A VISUALIZATION ON-DEMAND SERVER - A Service Oriented Architecture to Visually Explore
Large Data Collections
are neither central repository, nor evaluation. Some
lists exist, such as Many Eyes
or Visual Complex-
. The latter consists in an updated screenshot and
video repository: but there are no semantic taxonomy
providing automatic access to visualization.
2.1.1 Data Transformation Models
Our goal is to understand the visualization process
in general, regardless data types, task or application
domain. Then a model oriented analysis of the vi-
sualization has to be performed (Butler et al., 1993).
This approach has been commonly used, resulting in
many models. We focus on two major models from
two distinct fields that are Information Visualization
(Chi, 2000) and Scientific Visualization (Haber and
McNabb, 1990). See Figure 3.
Figure 3: From left to right, (Chi, 2000) and (Haber and
McNabb, 1990) models.
Both models consider the visualization process as
a data flow, conceptually cut into steps: they end up
with similar stages, but they are not complete enough
as they do not integrate interactions. We proposed a
slightly different model based on the previous ones,
but including interactions: we don’t consider the vi-
sualization process as a full data flow but merely as
a sequence of operations being independently per-
formed by means of actions. Our model decomposes
the Information Visualization process into 3 stages,
each coupled with user actions (see Figure 4). In-
teractions can be either manual (user’s action), either
semi-manual (user’s action triggers a system reaction)
or automatic (system’s action).
The model description is as follow:
Extraction: is a step considering any perspective of
visualization such as semantic (RDF), physical
Figure 4: Holistic data transformation model including both
data transformation processes and interactions.
(matrix, directories) or even social, which extract
different points of view from the data.
Selection succeeds in reducing the dataset by se-
lecting (SQL, SPARQL), aggregating and project-
ing in an appropriate manner, freeing the user
from a potential overload.
Layout: is to give a spatial attribute to abstract data
structure such as graph, lists or multidimensional
sets. The layout can be made out of 2D maps but
can also be a 3D model.
Organization is an action that will change the lay-
out attribute of data.
Render: is to transform abstract data layout into im-
ages or 3D models.
Filtering means to post-process (using image
analysis techniques such as Gaussian blur or
Laplacian filter) to provide a pre-processed visual
result (Vuillemot and Peralta, 2008). These math-
ematical computations based on 2D-signal trans-
formations help, for instance, users to fade details
or highlight contours.
The cut in layers helps to identify and describe
each technique, that can now be seen as independent
modular programs.
2.2 Service Oriented Architectures
Service Oriented Architecture (SOA) is an architec-
ture style that aims at reorganizing business processes
into loosely coupled packages. The packages are dis-
tributed into modules reachable over a network. The
simplicity of use and universality of access makes it a
design style helping the reuse of self-describing com-
ponents. The components are interfaced as individual
ICEIS 2008 - International Conference on Enterprise Information Systems
entities with the propensity to build applications very
quickly. Services goal is to allow functionality to be
stuck together to form ad-hoc applications reusing ex-
isting software service. The capabilities of adaptabil-
ity and evolution are high by adding a new service,
reducing developing costs and a quick deployment.
Service are software reachable over a network
independently from the underlying implementation.
Languages describing them are:
Interfaces are published in the WSDL file (XML-
Based Document that describes how to commu-
nicate with the Web Service) (Christensen et al.,
Service repositories store WSDL files and are us-
ing UDDI (Universal Description, Discovery and
Integration) helping to match with user’s needs.
(UDDI, 2000)
Clients having retrieved relevant interface will
contact the service provider using SOAP (SOAP,
Service composition (Agarwal et al., 2005) en-
ables resources to be merged and responds more
quickly and cost-effectively to changing market con-
ditions. Service helps not having license/software
distribution issues, and can reduce distribution costs,
piracy and reverse engineering.
2.3 Visualization Service
The challenge of generically benefiting from visu-
alization techniques has already been faced through
many researches carried out in various disciplines. As
far as we know, the closest and most prolific field
is Scientific Visualization, providing related architec-
tures. (Wood et al., 1996) introduced a client/server
communication based on a simple reference model to
perform data visualization. The visualization is done
over the web, and focuses only on a specific data
type which are plots. (Bonneau et al., 2005) intro-
duces a modular visualization environments enabling
users to change the data pipeline. A GUI interface
is described, where dynamic change of the visual-
ization pipeline can be done by users. Finally (Bla-
zona, 2007) is a Scientific Visualization system offer-
ing web services facilities, but which is not based on a
model. Thus visualization is seen as a whole process
which steps can’t be isolated.
Table 1: Characteristics of InfoVis and SOA.
Type/Field InfoVis SOA
Location Local Distributed
Coupling Tight Loose
Flexibility Bundle Package
Messages File format Messages
Communication None Protocol
The key idea of the Visualization On-Demand (Vi-
zOD) architecture is 1) to separate into indepen-
dent modules the visualization process according to
our model (as seen on Figure 4) and 2) distribute
processes, regardless the data being studied. This
modular approach has to be combined according to
a strategy, taking into account both technical con-
straints and users’ preferences.
Using on-demand paradigm means we consider
the rendered data stream as a media, such as movies
with Video On-Demand (VoD) that can be chosen ac-
cording to a user action. Conceptually, this expression
is well-fitted since we consider the visualization as a
stream of existing knowledge, just with another per-
spective. Technically, it holds many identical prop-
erties as VoD, such as cache, replication and perfor-
3.1 Architecture
The aim is to make visualizations technique as black
boxes, with a focus on interfaces and communication
3.1.1 Processes Subdivision
Subdividing means making smaller parts (called busi-
ness processes) of a larger system, operating inde-
pendently. Each business process has properties (de-
scribed in a UDDI repository) such as a description
including the task it achieves, performance and com-
plexity. These information will be useful to respect
users Quality of Service constraints.
While communication among the processes be-
comes asynchronous, modules keep interdependence
constraints. For instance a change in the layout will
trigger a new render. A color change in the graph de-
piction will leave an identical layout, but here again
render will have to be regenerated. We used as ex-
change protocol among the modules existing interme-
diate data transformation. For instance, the extraction
ANATOMY OF A VISUALIZATION ON-DEMAND SERVER - A Service Oriented Architecture to Visually Explore
Large Data Collections
module will communicate a graph structure to the lay-
out process as it would be done in a monolithic archi-
tecture. In other words, external interface mirror in-
ternal temporary data residing in computer memory.
The language choice turned out to be XML-like to
wrap messages as showed on Figure 5. Then an addi-
tional SOAP communication protocol layer is added.
Figure 5: Modules are interfaced with specific languages,
encapsulated in SOAP messages.
3.1.2 Processes Distribution
Business processes can be located at any places, and
reachable through a service repository. Process distri-
bution means that some processes may remain local
to users (e.g. because of privacy issues) while other
may be distributed and reachable other a network (e.g.
because they require lots of computing resources).
Figure 6: (a) strategy is to use remote layout process (b)
strategy is to keep local interactions only.
There exists many distribution strategies that can
be complex and dynamic (changing over time, ser-
vice availability or service load). Two examples are
described on Figure 6: (a) shows that user hosts the
dataset and transfer only data structure to perform a
layout process on it. This is a case where the layout
needs lots of computing power (b) shows an opposite
strategy, where user selects only the data and interacts
with the render only. This is a case where a dataset is
shared and reached through a local interactive envi-
3.2 Strategy of Access
A strategy is a common way of combining small steps
to tackle a bigger problem. A step will be regarded as
a group of processes, which are selected and glued
together to solve a task. Assembly of steps will be
done as close as the way the mind is working and
will be called patterns. These patterns are following
common orchestration that have been subject of study,
such as Schneiderman’s Overview Zoom and Details-
On-Demand (OZD) (Shneiderman, 1996).
3.3 Personalization
While a company is an organization having the same
focus, individuals have specificities to be considered.
Even if every task or context may be different, there
exists a visual knowledge about data structures (e.g.
tree-like structures similar as Figure 8) that are com-
mon to every kind of information. Learning how to
understand and master this knowledge requires time,
but once done it can result in time savings by cutting
delays to visually master new datasets. Thus, user’s
visual habits and preferences have to be identified and
Personalization (Brusilovsky et al., 2007) is a way
of taking user’s preferences into account. Researches
have been carried out in such direction as in informa-
tion retrieval systems, in order to reduce datasets ac-
cording to user’s explicit or implicit preferences. Ex-
plicit preferences are user selections and configura-
tions operations, such as local environment choice or
service choice. Implicit preferences are user’s history
or any typical behavior registered in a non-intrusive
way. For instance, if a specific service or group of
services are invoked many times, they will be consid-
ered as a preference, even if no question has ever been
asked to the user (Eirinaki and Vazirgiannis, 2003).
Preferences can also vary from short to long term
interest. Short term preferences are edge colors, any
encoded knowledge (such as symbols) and filters of
the render which aims at solving a task in a specific
context. Middle term preference is the data layout
which can’t vary (whereas colors can) in order not
to disturb user’s mental model, which is his internal
vision of a virtual scene. Finally, a long term inter-
est concerns the interactive environment in which are
integrated visualizations.
Preferences will be stored in a User Visual Profile
and will be available regardless user’s location.
ICEIS 2008 - International Conference on Enterprise Information Systems
To validate our architecture, we developed a first
VizOD prototype using existing visualization tech-
niques and interactive environments. Advanced tech-
nical details are available in (Vuillemot and Peralta,
2008). Our approach was to implement the strategy
described in Figure 6 (b) which Use Case is available
Figure 7.
Figure 7: Prototype Use Case.
The dataset is a movie database
. The data ex-
traction consists in performing queries, selecting 1)
users and 2) for each users their rated movies. That
selection is done by means of a web interface allow-
ing SQL-like queries. The extracted result consists
in a tree-like structure where the root is an artificial
node connecting all users with their rated movies as
leaves. The graph layout techniques used was origi-
nally aimed to display protein networks (Adai et al.,
2004). Other graph visualization tools exist such as
(Auber, 2003). The render step results in an image
with annotated data (containing details about movies),
provided in a separate file structured in a XML-like for-
mat, KML (Keyhole Markup Language). The result is
the picture displayed on Figure 8.
The image looks intriguing, but lacks of efficiency
since it holds only structural information: it has to be
included in a user’s interactive environment and be
more detailed.
We selected Google earth (GE) as an interactive
environment (with all geo-spatial features disabled).
A screenshot is available Figure 9. GE is installed and
run by the client, and connects to VizOD by mean of
HTTP requests. Using HTTP helps to keep away fire-
wall issues or any complicated network configuration.
VizOD will seamlessly map onto GE’s 3D sphere an
image according to user’s altitude and angle of view,
following (Shneiderman, 1996)’s recommendations.
Figure 8: A tree-like layout visualization of a query result.
The result is a 3-layered multi-scale strategy com-
bining external business processes, and resulting (by
decreasing altitude) in an overview layer, zoom layer
and details layer. The overview layer aims at showing
global trends, then it will be connected to a blurring
render service, to remove details and raise trends. The
service interfaces Gimp
used in command line. The
zoom layer remains the original image. The detail
layer is the original image augmented with details on
top of it (included in the KML file). Bandwidth usage
has been minimized with that detail layer, by keep-
ing the zoom image locally and only requesting and
adding light KML data on top of it. The KML is con-
verted to additional vectorial graphics (lines, captions,
..) by GE.
It takes about 59s (on an ordinary server) to gen-
erate a single image. The layout process is the greed-
iest one, and generating or blurring images involves
nearly no extra time or resource use cost.
Figure 9: Image rendered mapped onto a 3D-sphere with
details on top of it.
Results, issued from experiments, was that GE is
a good metaphor since it implements a well-known
GNU Image Manipulation Program available at:
ANATOMY OF A VISUALIZATION ON-DEMAND SERVER - A Service Oriented Architecture to Visually Explore
Large Data Collections
object that is earth. And users already used GE for
other purpose: then we managed to minimize the en-
vironment mastering phase by reusing an existing and
widespread tool. Usability was excellent since we re-
used a powerful environment, which remains very re-
active to every gestures and moves from users, even if
images were not fully loaded or updated yet. Thanks
to our VizOD approach, many visualization strate-
gies can be adopted, while the client stays focused
to a very same interface. The dataset can even to-
tally change without absolutely no interface change.
Adding new features on VizOD will be transparent
for users. Finally users appreciated to use an attrac-
tive means that is a 3D sphere (similar to the iPod
effect where the appealing wheel attracts users).
In this section we discuss applications of our architec-
Company benefits issued from using a VizOD ar-
chitecture will come by outsourcing visualization to
experts, providing software as services with better
support and piracy prevention. The information sys-
tem rationalization will go further by centralizing
computing power at the same remote place and leav-
ing end-users with lightweight heterogeneous termi-
nals. The software maintenance routine is not on site
any more but on the VizOD servers, holding business
processes, which can be numerous allowing diversity
and backups alternatives. New economical models
will appear for producers.
Privacy and security issues are a big concern in
our approach. Regarding the steps of our model one
can see that rendered images prevent reverse engi-
neering process. Furthermore, extracting data struc-
tures only will give structural information and pre-
venting details being visible: quantity of information
is given, not quality. Finally, a service approach pre-
vents implementation details to be visible. However,
we keep SOA related issues, such as messages ex-
change that is prone to attacks.
User’s needs have to be considered globally, with
imperfections, such as short attention spans. Memo-
ries are also important aspects to deal with, especially
with data masses and in the case information is avail-
able in streams which are not stored: user’s full atten-
tion is then required. The focus can be on visualizing
updates, data growth, rather than content itself: the
change or the behavior becomes as important as the
instant content. The scalability and adaptability of the
VizOD architecture is crucial.
Services Mashup interfaces are new way for users
to compose services to build up and share new ones.
But is does not include yet extra process such as data
layout and render. A future work is to implement a Vi-
sualization Mashup Interface to cope with the lack of
semantic and integration visualization service repos-
itory. User needs tools in new era where web-users
have become actors using web interfaces.
New design process and product life cycle will
emerge. Programmers are not constrained, then they
will keep their own programming habits. A piecewise
conception process can be set up, with progressive
features. Other Research communities are addressed
such as cognitivists in order to observe users behav-
ior, interface designer to re-think the way to design
and evaluate. New actors such as artists can now fully
take part to the design process by including artistic
among one or many navigation step.
In this paper we introduced the outlines of a visu-
alization on-demand architecture (VizOD). We first
proposed a holistic data transformation model, con-
sidering both visualizations and interactions. Every
step of the model are considered as business processes
that can either remain local or be distributed. Such an
approach allows end-users to benefit and personalize
visualization services, and couple it into their local
interactive environments.
A present day result is a prototype resulting from
an assembly of existing tools and showing that even
non visualization-dedicated tools (such as graph ma-
nipulation libraries) can quickly result in an innovat-
ing application, following VizOD’s specifications, in
a context of affordable systems and non-expensive
The missing masterpiece in our approach is a way
to encapsulate processes to make them automatically
discoverable and usable. There is also a semantic gap
between user’s task needs which are expressed in nat-
ural language and tasks the machine already knows
how to achieve. To tackle this issue, our next step is
building on line communities that will fertilize best
practices or novel uses. We will also focus on user
generated data (e.g. traces of use) that have to be
structured, filtered and sorted. More generally, a sta-
ble on line framework has to emerge and be sustain-
able over time, and then lessons will be learn by wide-
spread use.
ICEIS 2008 - International Conference on Enterprise Information Systems
Adai, A. T., Date, S. V., Wieland, S., and Marcotte, E. M.
(2004). Lgl: creating a map of protein function with
an algorithm for visualizing very large biological net-
works. J Mol Biol, 340(1):179–190.
Agarwal, V., Dasgupta, K., Karnik, N., Kumar, A., Kundu,
A., Mittal, S., and Srivastava, B. (2005). A service
creation environment based on end to end composi-
tion of web services. In WWW ’05: Proceedings of
the 14th international conference on World Wide Web,
pages 128–137, New York, NY, USA. ACM.
Auber, D. (2003). Tulip : A huge graph visualisation frame-
work. In Mutzel, P. and J
unger, M., editors, Graph
Drawing Softwares, Mathematics and Visualization,
pages 105–126. Springer-Verlag.
Blazona, Bojan; Mihajlovic, Z. (25-28 June 2007). Visu-
alization service based on web services. Information
Technology Interfaces, 2007. ITI 2007. 29th Interna-
tional Conference on, pages 673–678.
Bonneau, G.-P., Ertl, T., and Nielson, G. M. (2005).
Scientific Visualization: The Visual Extraction of
Knowledge from Data. Mathematics+Visualization.
Brusilovsky, P., Kobsa, A., and Nejdl, W. E. (2007). The
Adaptive Web.
Butler, D. M., Almond, J. C., Bergeron, R. D., Brodlie,
K. W., and Haber, R. B. (1993). Visualization ref-
erence models. In VIS ’93: Proceedings of the 4th
conference on Visualization ’93, pages 337–342.
Chi, E. H. (2000). A taxonomy of visualization techniques
using the data state reference model. In INFOVIS,
pages 69–76.
Christensen, E., Curbera, F., Meredith, G., and Weer-
awarana, S. (2001). Web Services Description Lan-
guage (WSDL) 1.1. Technical report, W3C Note,
Eirinaki, M. and Vazirgiannis, M. (2003). Web mining for
web personalization. ACM Trans. Inter. Tech., 3(1):1–
Haber, R. and McNabb, D. (1990). Visualization idioms: A
conceptual model for scientific visualization systems.
In Visualization in Scientific Computing, pages 74–93,
G.M. Nielson, B. Shriver, and L.J. Rosenblum, eds,
CS Press Los Alamitos, Calif. IEEE.
Michelle, Woodruff, A., and Kuchinsky, A. (2000). Guide-
lines for using multiple views in information visual-
ization. In Advanced Visual Interfaces, pages 110–
Shneiderman, B. (1996). The eyes have it: A task by data
type taxonomy for information visualizations. In VL
’96: Proceedings of the 1996 IEEE Symposium on
Visual Languages, page 336, Washington, DC, USA.
IEEE Computer Society.
Shneiderman, B. and Plaisant, C. (2004). Designing the
User Interface : Strategies for Effective Human-
Computer Interaction (4th Edition). Addison Wesley.
SOAP (2000). Simple object access protocol (soap 1.1).
UDDI (2000). Universal description, discovery and inte-
gration, version 3. OASIS, Billerica, Mass., 2000;
Vuillemot, R. and Peralta, V. (2008). From Beautiful to Use-
ful: A Multi-Scale Visualization of Users Movie Rat-
ings. Technical Report RR-LIRIS-2008-001, LIRIS
UMR 5205 CNRS/INSA Lyon.
Wood, J., Brodlie, K., and Wright, H. (1996). Visualiza-
tion over the world wide web and its application to
environmental data. In VIS ’96: Proceedings of the
7th conference on Visualization ’96, pages 81–ff., Los
Alamitos, CA, USA. IEEE Computer Society Press.
ANATOMY OF A VISUALIZATION ON-DEMAND SERVER - A Service Oriented Architecture to Visually Explore
Large Data Collections