Virtual Reality Techniques for 3D Data-Warehouse Exploration
Hamza Hamdi, Eulalie Verhulst and Paul Richard
Laboratoire Angevin de Recherche en Ing
enierie des Syst
emes (LARIS - EA 7315),
e d’Angers, Angers, France
Virtual Environments, Interaction Techniques, Navigation, Data Warehouse, Human Performance.
This paper focuses on the evaluation of virtual reality (VR) interaction techniques for exploration of data ware-
house (DW). The experimental DW involves hierarchical levels and contains information about customers
profiles and related purchase items. A user study has been carried out to compare two navigation and selection
techniques. Sixteen volunteers were instructed to explore the DW and look for information using the inter-
action techniques, involving either a single Wiimote
(monomanual) or both Wiimote
and Nunchuck
(bimanual). Results indicated that the bimanual interaction technique is more efficient in terms of speed and
error rate. Moreover, most of the participants preferred the bimanual interaction technique and found it more
appropriate for the exploration task. We also observed that males were faster and made less errors than females
for both interaction techniques.
Worldwide corporations are mining their data to learn
about client purchasing patterns, fraud, credit appli-
cations and health care outcome analysis. It ap-
pears that the worldwide business intelligence and
data warehousing market had about a 60% year-over-
year growth. As one expects, there has been a surge in
the number of applications that provide DW creation,
management and mining. A DW is a subject-oriented,
integrated, time-variant and non-volatile collection of
data in support of management’s decision making
process (Han and Kamber, 2000).
During the last decade, significant research efforts
have been devoted towards facilitating access to in-
formation. In this context, typical relational database
systems were optimized for query processing and on-
line transaction processing (OLTP), which minimizes
the time needed for systematic daily operations of an
organization (Sawant et al., 2000; Ammoura et al.,
2001). VR techniques were also proposed to immerse
experts in 3D DW. This approach relies heavily on the
human abilities to explore, perceive, and process 3D
information. In this context, different VR setup and
application have been developed (Vald
es, 2005; Ogi
et al., 2009; Nagel et al., 2008). Most of these in-
volved complex systems and user interfaces such as
the CAVE
(Cruz-Neira et al., 1992). With the con-
tinuous improvements in computer technology and
video games, it is now possible with almost standard
PCs and low-cost interaction devices such as the Nin-
tendo Wiimote
to explore 3D data sets. However,
usability studies and human performance evaluation
still need to be carried out in order to reach intuitive
and efficient interaction techniques.
In this paper, we report on a user study aimed to
compare two different interaction techniques for the
exploration of a data warehouse with specific graphics
encoding. Sixteen volunteers were instructed to look
for information using an interaction technique in-
volving either a Wiimote
(monomanual technique)
or both a Wiimote
and a Nunchuck
In the next section, we give a short survey about
VR techniques used in the context of Visual Data
Mining (VDM). Then we describe some works about
3D interaction techniques in Virtual Environments
(VEs). In section 3, we describe the experimental
data set and the proposed graphic encoding of these
data. Section 4 presents the proposed interaction
techniques based on the Nintendo Wiimote
. Section 5 is dedicated to the experi-
ment and the results analysis. The paper ends by a
conclusion and gives some tracks for future work.
Hamdi H., Verhulst E. and Richard P.
Virtual Reality Techniques for 3D Data-Warehouse Exploration.
DOI: 10.5220/0006130400750083
In Proceedings of the 12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2017), pages 75-83
ISBN: 978-989-758-229-5
2017 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
2.1 Visual Data Mining
Visual data mining (VDM) aims to integrate the hu-
man in the data exploration process, applying his per-
ceptual abilities to the large data sets available in to-
day’s computer systems. The main idea of VDM is
to present the data set in some visual form, allowing
the human to get into the data, draw conclusions, and
directly interact with the data. VDM techniques have
proven to be of high value in exploratory data analy-
sis and they also have a high potential for exploring
large databases. According to (Wong and Bergeron,
1997) the exploratory analysis of data and VDM are
not only a set of tools but also a philosophical manner
to approach the problem of knowledge discovery.
VDM methods have been implemented in VR in
several occasions as well as traditional methods for
data exploration (Symanzik et al., 1997; Wegman and
Symanzik, 2002). In 1999, a research project called
3D Visual Data Mining (3DVDM) was initiated at
the VR Media Lab at Aalborg University to study
how VR may be used in VDM (Nagel et al., 2001;
Granum and Musaeus, 2002). Among the facilities
of the VR Media Lab are a 3D Power Wall, a 160
degree Panorama, a 6-sided CAVE, and a 16 proces-
sor SGI Onyx2. Another project called DIVE-ON
(Data mining in an Immersed Virtual Environment
Over a Network), uses advance in VR, databases, and
distributed computing to experiment with a new ap-
proach to VDM. For example, the DWs generated by
DIVE-ON were N-dimensional data cubes (Ammoura
et al., 2001).
2.2 3D Interaction Techniques
3D interaction techniques are generally classified as
follows (Mine, 1995; Hand, 1997; Zeleznik et al.,
1997): selection, manipulation, navigation and appli-
cation control. Navigation is composed of two tasks:
travelling and way-finding (Bowman et al., 2001),
where travelling represents the main component of
navigation and refers to the physical displacement
from a place to another one. Way-finding corresponds
to the cognitive component of navigation by allowing
the users to be located in the VEs and to choose a tra-
jectory for displacement. Both aspects of navigation
are crucial for efficient exploration of 3D DW.
Several studies aimed to develop techniques for
specific tasks and applications. (Bowman et al., 1997)
suggested a framework for evaluating the quality of
interaction techniques for specific tasks in VEs. Re-
sults indicated that pointing techniques are advanta-
geous relative to gaze-directed steering techniques for
a relative motion task. Moreover, they observed that
motion techniques which instantly teleport users to
new locations are correlated with increased user dis-
orientation. Some hand directed motion techniques
have also been proposed for navigation. The position
and orientation of the hand determines the direction
of motion through the VEs. Wii devices have been
adopted by a number of researchers for a wide vari-
ety of applications (Schlomer et al., 2008). Gener-
ally, in case of using The Wiimote
for navigation,
rotation angles such as pitch, yaw, and roll informa-
tion are used. For example, (Duran et al., 2009) used
the Wiimote
for controlling wheelchair using pitch
and yaw movements. (Fikkert et al., 2009; Fikkert
et al., 2010) proposed interaction techniques using the
and the Wii Balance Board
. Both in-
put devices were used to navigate a maze application.
(Yamaguchi et al., 2011) developed a 3D interaction
technique to explore Google Earth using the Nintendo
Wii devices. The Wiimote
was used for zooming
and steering and the Balance Board
was used for
walking. The authors tested operation workload for 9
different threshold angle combinations. They found a
most low workload threshold angle combination of 45
degrees (for zooming out) /-15 degrees (for zooming
in) and of 30 degrees (for steering right) /-30 degrees
(for steering left). Some more recent approaches have
been proposed for navigation in VEs (Fajnerova et al.,
2015; Gaona et al., 2016; Christou et al., 2016).
The aim of the graphic encoding is the rewriting of
the data in the form of graphic objects by associ-
ating each variable in the data with a graphic ones
(position, length, surface, color, luminosity, satura-
tion, form, texture, etc.). Graphic objects can be from
zero to three dimensions, i.e. a point, a line, a sur-
face, or a volume. Evolution of the objects in time
can introduce an additional dimension. Several au-
thors worked with the classification of the graphic
encodings so as to determine those which are most
effective according to the data to represent. For
the statistical graphs (groups of dots, diagrams, etc),
there are works from(Cleveland and McGill, 1984),
and(Wilkinson, 1999).
3.1 Description of the Database
For the experiment, we modelled a simple man-
agement of client relationship (CRM) database that
HUCAPP 2017 - International Conference on Human Computer Interaction Theory and Applications
gather information enabling the describtion and char-
acterization of customers purchase. This database is
primarily characterized by:
the Customer table that is characterized by the
identifier, first name, familly name, age, sex, mar-
ital status and the number of credits;
the Product, table where each product is identi-
fied by a single code and is indicated by the word-
ing, the category of product (fruits, vegetables,
drinks), the unit price and stock.
This database is described under the Microsoft
Access basic management system, and include 200
customers. The Figure 1 illustrates the selected
database using the relational model.
Figure 1: Model of data design.
3.2 Graphic Encoding
Each element of the database was translated to a
graphic representation (position and size). The pro-
posed graphic encoding of each customer is illus-
trated in Figure 2. The height of the cone repre-
sents the number of purchased items, while the ra-
dius of the sphere represents the percentage of the
expenditure relativelly to the other customers. The
color of spheres and cones were randomly selected
in order to easily distinguish each customer. More
precisely, a large cone posed with the lower part of
a sphere represents a customer whose expenditure is
high, while a small cone posed with the lower part of
a sphere represents an encoding of a low expenditure
customer. Moreover, complementary text labels are
posted above each object to give the first name and
familly name of the customer.
Each customer has been placed on the right (fe-
males) or left (males) side of the VE according to
his/her gender (Fig. 3 (a)). As we can see, this
graphic encoding strongly highlights the most active
customers. Similarly, the customers have been placed
at different depth according to their age, as illus-
trated in Table 1. The database contains all informa-
tion about customers’ purchases history (last twelves
Figure 2: Graphic encoding of each customer.
months only) which have been classified according
to their categories (cloths, fruits, meat vegetable, yo-
ghurts, fish, cheese, drinks). These data are visualized
using three-dimensional histograms representing the
amount of each product versus the month they have
been purchased. The histograms are positioned simi-
larly to the shelves of a supermarket (Fig. 3 (b)).
Table 1: Custumers segmentation according to their age.
4.1 Interaction Modelling
We proposed two interaction modes for the explo-
ration of the data-warehouse: (1) the navigation mode
and (2) the selection mode. The navigation mode al-
lows the user to navigate and explore both the VE
containing the customers representation and each VE
containing the customers data. This appoach is illus-
trated by the hierarchical model presented in Figure 4.
The Translate function allows the user to move the
camera (viewpoint) along the lateral axis and/or
the depth axis;
The Zoom function allows the user to zoom on a
selected customer, whatever his/her distance from
the user;
The Rotate function allows the user to rotate the
camera relative to the vertical axis (steering).
Virtual Reality Techniques for 3D Data-Warehouse Exploration
Figure 3: Illustration of the 3D database: customers visual-
ization (a), purchased items of a given customers (b).
The selection mode is split in three sub-modes includ-
ing Pre-selection, Entry, and Exit. It allows the user
to select a given customer at any time and at any dis-
tance from him/her. This mode is illustrated by the
hierarchical model presented in the Figure 5.
The Pre-selection function enables an automatic
zoom on a selected customer in order to quickly
and easily get his/her personal information;
The Entry function enables the user to be tele-
ported in the VE containing the selected customer
personnal data;
The Exit function enables the user to get back to
the main VE. This function could be activated at
any time and anywhere in the VE containing the
customer personnal data and do not require any
selection process.
4.2 Implementation of Interaction
In order to propose a low-cost system, we im-
plemented the interaction technique using the
and the Nunchuck
. The Wiimote
(Fig. 6) has the capability to track the user’s hand ori-
entation along two degrees of freedom (pitch and roll)
using inertial sensors (accelerometers / gravimeters).
In addition, the Wiimote
has 12 buttons which
could be used to trigger events. The Wiimote
also equiped with an infrared emitter which could be
used to track yaw movements. The Wiimote
municates with the computer using Bluetooth wire-
less communication an could be connected with the
in order to add a second set of 3 ac-
celerometers along with 2 trigger-style buttons and
Figure 4: Hierarchical model for the navigation mode.
Figure 5: Hierarchical model of the selection mode.
an analog joystick. therefore, the combination of the
may be used as a bimanual
user input.
4.2.1 Mono-manual Interaction Technique
For this approach, the Wiimote
was used to carry
out forward and backward movements (pitch) and
steering movements (roll). It was also used to switch
between the navigation and the selection modes (B
button). This approach allows simultaneous control
of translation and rotation of the virtual camera. In the
selection mode, the mouse cursor was also moved us-
ing the pitch and roll movements preventing the use of
the infrared emmiter/receiver. To select a given cus-
tomer and trigger the teleportation of the user in the
VE containig his/her peronnal data (customer world),
the A button of Wiimote
was used.
HUCAPP 2017 - International Conference on Human Computer Interaction Theory and Applications
Figure 6: Rotational degrees of freedom (pitch, roll and
yaw) provided by the Wiimote
4.2.2 Bi-manual Interaction Technique
The proposed bi-manual interaction technique use
both the Wiimote
and the Nunchuck
. The for-
mer device is used for only for customer selection
and trigger the teleportation and the come back of the
user in the main VE. The latter device is exclusivelly
used for navigation and therefore to control the for-
ward/backward (Fig. 7 (a)) and steering (Fig. 7 (b))
movements of the virtual camera. This is done using
the joystick of the Nunchuck
. As in the previous
navigation technique, this approach allows simulta-
neous control of translation and rotation of the virtual
(a) (b)
Figure 7: Control of forward/backward (a) and steering (b)
movements using the Nunchuck
5.1 Task
Each participant was instructed to explore the cus-
tomers world and find out the three best customers
(higher cones), among one hundred customers. Then,
they had to select each of these customers and explore
their personal data (customer world) to discover the
three most purchased items. This procedure is illus-
trated in Figure 8 using a finite state machine. The
task ends as the participant felt he/she collected the
requested information and get back to the main world.
Figure 8: State machine model of the exploration task.
Whatever his/her location, a click on a given cus-
tomer (using the A button of the Wiimote
) en-
ables a teleportation of the virtual camera in front of
him/her. This allows the participants to read the tex-
tual information about the customer’s name and give
him/her the possibility to be teleported in the cus-
tomers world. Thus, the participant may then decide
to enter in the customer world (using the B button of
the Wiimote
), or to get back to its initial position
and continue the exploration of main world. In the
customer’s world, the participants can navigate us-
ing the same interaction technique and find out the
requested information (three most purchased items).
The participants can then activate the B button of the
to get back to their location in the main
world (in front of the previouly selected customer).
5.2 Design and Procedure
Sixteen participants (8 males and 8 females) were
divided into two groups (G1 and G2). They were
aged between 22 and 35 years old (72.72% are right-
handed and 27.27% are left-handed) and had normal
or corrected-to-normal vision capabilities. The exper-
iment was conducted according to the two following
C1: navigation and selection using Wiimote
only (Fig. 9 a);
C2: navigation using the Nunchuck
, selection
using the Wiimote
(Fig. 9 b).
Four different VEs (D1, D2, D3 and D4) each cor-
responding to a different database were defined. The
participants of the group G1 start the exploration of
D1 and D2 by using Wiimote
only (condition C1).
Then, they explore D3 and D4 by using Wiimote
Virtual Reality Techniques for 3D Data-Warehouse Exploration
and Nunchuck
(condition C2). Similarly, the par-
ticipants of the group G2 start the exploration of D1
and D2 by using Wiimote
and Nunchuck
dition C2), and they also explore D3 and D4 by using
only (condition C1). In order to facilitate
the comprehension of the experience, we give a short
description of the task and allow each participant to
get acquainted with the system and perform in both
conditions (C1 and C2).
Figure 9: A participant exploring the 3D datawarehouse us-
ing: (a) the Wiimote
only (condition C1), and (b) both
the Wiimote
and the Nunchuck
(condition C2).
Each participant was placed in front of a large
rear-projected stereoscopic screen (2x2.5 m) as illus-
trated in Figure 9, and equiped with passive (polar-
ized) glasses. A Full HD Optoma HD142X (1080p)
projector was used for displaying the images. The
participants were instructed to start the task using ei-
ther the Wiimote
(C1) or both the Wiimote
(C2) according to the group they be-
long to. At the end of the experiment, we gave each
participant a questionnaire in order to get subjective
data about the proposed interaction techniques and the
graphic encoding of the data warehouse.
5.3 Collected Data
We recorded the task completion time and the number
of errors for each single trial for each database (D1,
D2, D3, and D4). The errors consisted in the wrong
selection of the three best customers or the most three
purchased items. In order to examine participant’s be-
havior during the task, we recorded the paths for each
single trial.
5.4 Results
In this section we present the results of the experi-
mental study. The collected data have been analyzed
through a one-way repeated ANOVA. The descrip-
tion of the results is based on three criteria: (1) task
completion time, (2) average number of errors (num-
ber of best custumers not selected or number of the
three most purchased item not selected), and (3) par-
ticipant’s paths, which have been recorded in order to
analyse his/her behaviour and strategy or any difficul-
ties concerning the task.
5.4.1 Effect of Navigation Technique
Task Completion Time. The ANOVA revealed a
significant effect of the navigation technique on the
task completion time (F(1, 23) = 20.19; p < 0.05).
Average completion time for the condition C1 and C2
condition were respectively about 364.3 Sec. (SD :
36.30) and 300.7 Sec. (SD : 29.75). This result is
illustrated in Figure 10.
Figure 10: Task completion time (Sec.) vs. condition C1
(monomanual) and C2 (bimanual).
Number of Errors. The average number of errors
for the condition C1 (monomanual) and C2 (biman-
ual) were 4.27 and 3.73 respectively. A statistical
HUCAPP 2017 - International Conference on Human Computer Interaction Theory and Applications
analysis (ANOVA) showed that the interaction tech-
nique, either monomanual or bimanual, has no signif-
icant effect on error rate.
5.4.2 Gender Effect
In the following paragraph we look at the gender
effect on participant’s performance such as the task
completion time and number of errors.
Task Completion Time. We observed a sig-
nificant gender effect on task completion time
(F(1, 23) = 42.73; p < 0.05). Average completion
time was about 266 Sec. (SD : 40.5) for males and
399 Sec. (SD : 53.9) for females (Fig. 11). This sug-
gest that females had more difficulties in exploring
the VEs and find out the requested information than
Figure 11: Task completion time (Sec.) vs. gender.
Number of Errors. The number of errors was on
average 3.14 for males, and 6.33 for females (Fig. 12).
Thus, females made more than twice number of errors
than males.
Figure 12: Number of errors vs. gender.
Navigation Paths. Figures 13 (a) and 13 (b) illus-
trate typical paths associated respectively with males
(a) (b)
Figure 13: Typical paths obtained by males (a) and females
and females. We observed that most females used no-
optimal and longer routes than males.
5.5 Subjective Data
Each participant were instructed to fill a questionnaire
and answer the following questions.
Question 1 Which interaction technique you feel the
most relevant for the task?
Question 2 How would you rate the level of diffi-
culty of the task (simple/difficult/very difficult)?
Results showed that many participants found the
bimanual interaction technique more relevant than the
monomanual one. Thus, 63.63% of them preferred
to perform the task using the Wiimote
with the Nunchuck
and 36.36% of them preferred
to perform the task with the Wiimote
only. Con-
cerning the level of difficulty of the task, many partic-
ipants (72%) found it simple and had no difficulty to
understand the meaning of the graphics.
5.6 Discussion
In this paper we compared mono and bimanual
interaction technique for the exploration of a data
warehouse. Participants used Wiimote
alone or
and Nunchuck
. The results revealed
that interaction technique has a significant effects
Virtual Reality Techniques for 3D Data-Warehouse Exploration
on the objective dependant variables (completion
time and error rate). We observed that the bimanual
interaction technique (Wiimote
and Nunchuck
led to better performance. In addition, most of partic-
ipant preferred this technique over the monomanual
one (Wiimote
These results confirm some previous results con-
cerning the efficiency of bimanual interaction tech-
nique for complex motor activities (Guiard, 1987). Of
course, any combinations of these devices would have
been possible like selection with the Nunchuck
navigation with the Wiimote
. However, the use of
the Wiimote
for selection and the Nunchuck
navigation is efficient. Indeed joysticks are easy to
use(Vera et al., 2007) and only the Nunchuck
has a
joystick. The Nunchuck
may have a better usabil-
ity than the Wiimote
. The observation of paths in
the cutomers world revealed that the Nunchuck
more appropriate for navigation than the Wiimote
(less saccadic trajectories). We have observed a sig-
nificant gender effect on task performance. The males
had much better performance in terms of task com-
pletion time and error rate than females with both
interaction techniques. It seem that this difference
come from the difficulties for females to navigate
and explore the 3D worlds(Vila et al., 2003). These
results allow us to state that 3D perception is rela-
tively low for females (Fig. 13)(Coluccia and Louse,
2004). Futhermore, these difficulties may have been
increased by the complexity of the task and the multi-
level hierachy of the data warehouse involving back-
and-forth transposition between the customer world
and the purchased items worlds.
In this paper, we investigated the use of VR tech-
niques for the exploration of a data warehouse. In this
context, we have proposed two interaction techniques
based on the Wiimote
and Nunchuck
. Volunteer
participants were instructed to explore a data ware-
house with specific graphics encoding and collect in-
formation, either using the Wiimote
only for sec-
tion and navigation (monomanual technique) or using
the Nunchuck
for navigation and the Wiimote
for selection (bi-manual technique). Results indicated
that the proposed bimanual interaction technique is
more efficient than the mono-manual one in terms
of completion time and errors rate. Moreover, most
participants preferred the bimanual interaction tech-
nique and find it easier and more appropriate for the
required task. Finally, we observed that males were
much faster and made less errors than females for
both interaction techniques. This confirm previous
results concerning the difficulties of females for navi-
gating in complex 3D worlds(Vila et al., 2003). In the
future work, we will develop and investigate interac-
tion and navigation techniques such as step-in place.
In addition, we will introduce haptic and multimodal
assistances to help the users to find information in the
data warehouse. We will also pay attention to the us-
ability of the interaction technique and users habits.
Agarwal, S., Agrawal, R., Deshpande, P., Gupta, A., F.,
J. N., Ramakrishnan, R., and Sarawagi, S. (1996). On
the computation of multidimensional aggregates. In
Proceedings of the 22
International Conference on
Very Large Data Bases, pages 506–521, San Fran-
cisco, CA, USA.
Ammoura, A., Zaane, O. R., and Ji, Y. (2001). Dive-on:
From databases to virtual reality. Crossroads, 7(3):4–
Bowman, D. A., Johnson, D. B., and Hodges, L. F. (2001).
Testbed evaluation of virtual environment interaction
techniques. Presence, 10(1):75–95.
Bowman, D. A., Koller, D., and Hodges, L. F. (1997).
Travel in immersive virtual environments: An evalua-
tion of viewpoint motion control techniques. Virtual
Reality Annual International Symposium, 215:45–52.
Christou, C., Tzanavari, A., Herakleous, K., and Poullis,
C. (2016). Navigation in virtual reality: Compar-
ison of gaze-directed and pointing motion control.
In 2016 18th Mediterranean Electrotechnical Confer-
ence (MELECON), pages 1–6.
Cleveland, W. S. and McGill, R. (1984). Graphical per-
ception: Theory, experimentation, and application to
the development of graphical methods. The American
Statistical Association, 79(387):531–554.
Coluccia, E. and Louse, G. (2004). Gender differences in
spatial orientation: A review. Journal of Environmen-
tal Psychology, 24(3):329 – 340.
Cruz-Neira, C., Sandin, D., DeFanti, T., Kenyon, R., and
Hart, J. (1992). The cave: audio visual experience au-
tomatic virtual environment. Communications of the
ACM, 35(6):65–72.
Duran, L., Fernandez-Carmona, M., Urdiales, C., Peula, J.,
and Sandoval, F. (2009). Conventional joystick vs. wi-
imote for holonomic wheelchair control. 5517:1153–
Fajnerova, I., Rodriguez, M., Spaniel, F., Horacek, J., Vl-
cek, K., Levcik, D., Stuchlik, A., and Brom, C. (2015).
Spatial navigation in virtual reality : from animal
models towards schizophrenia: Spatial cognition tests
based on animal research. In International Conference
on Virtual Rehabilitation (ICVR), pages 44–50.
HUCAPP 2017 - International Conference on Human Computer Interaction Theory and Applications
Fikkert, F., Hoeijmakers, N., van der Vet, P., and Nijholt, A.
(2009). Navigating a maze with balance board and wi-
imote. Intelligent Technologies for Interactive Enter-
tainment, Lecture Notes of the Institute for Computer
Sciences, Social Informatics and Telecommunications
Engineering, 9:187–192.
Fikkert, F., Hoeijmakers, N., van der Vet, P., and Nijholt,
A. (2010). Fun and efficiency of the wii balance in-
terface. International Journal of Arts and Technology
(IJART), 3(4):357–373.
Gaona, P. A. G., Moncunill, D. M., Gordillo, K., and
Crespo, R. G. (2016). Navigation and visualiza-
tion of knowledge organization systems using virtual
reality glasses. IEEE Latin America Transactions,
Granum, E. and Musaeus, P. (2002). Constructing vir-
tual environments for visual explorers, pages 112–
138. Springer-Verlag, London, UK.
Gray, J., Chaudhuri, S., Bosworth, A., Layman, A., Re-
ichart, D., Venkatrao, M., Pellow, F., and Pirahesh, H.
(1997). A relational aggregation operator generalizing
group-by cross-tab, and sub-totals. pages 152–159.
IEEE International Conference on Data Engineering.
Guiard, Y. (1987). Asymmetric division of labor in hu-
man skilled bimanual action: The kinematic chain as
a model. Journal of Motor Behavior, 19:486–517.
Han, J. and Kamber, M. (2000). Data Mining: Con-
cepts and Techniques (The Morgan Kaufmann Series
in Data Management Systems). Morgan Kaufmann.
Hand, C. (1997). A survey of 3d interaction techniques.
Computer Graphics Forum, 16:269281.
Mine, M. R. (1995). Isaac: A virtual environment tool for
the interactive construction of virtual worlds. Techni-
cal report, University of North California, Chapel Hill
Nagel, H. R., Granum, E., Bovbjerg, S., and Vittrup, M.
(2008). chapter immersive visual data mining the
3dvdm approach. Visual Data Mining, pages 281–
Nagel, H. R., Granum, E., and Musaeus, P. (2001). Methods
for visual mining of data in virtual reality. PKDD:
International Workshop on Visual Data Mining.
Nelson, L., Cook, D., and Cruz-Neira, C. (1999). Xgobi
vs the c2: Results of an experiment comparing data
visualization in a 3-d immersive virtual reality envi-
ronment with a 2-d workstation display. In Computa-
tional Statistics: Special Issue on Interactive Graph-
ical Data Analysis, pages 39–51. Communications of
the ACM.
Ogi, T., Tateyama, Y., and S, S. (2009). Methods for visual
mining of data in virtual reality. In 3
Conference on Virtual and Mixed Reality, pages 13–
Sawant, N., Scharver, C., Leigh, J., Johnson, A., Rein-
hart, G., Creel, E., Batchu, S., Bailey, S., and Gross-
man, R. (2000). The tele-immersive data explorer :
A distributed architecture for collaborative interactive
visualization of large data-sets. In 4
Immersive Projection Technology Workshop, Ames,
Iowa, USA.
Schlomer, T., Poppinga, B., Henze, S., and Boll, S. (2008).
Gesture recognition with a wii controller. In Tangible
and Embedded Interaction.
Symanzik, J., Cook, D., Kohlmeyer, B., Lechner, U., and
Cruz-Neira, C. (1997). Dynamic statistical graphics in
the c2 virtual reality environment. Computing Science
and Statistics, 29:41–47.
es, J. J. (2005). Visual data mining of astronomic data
with virtual reality spaces, understanding the underly-
ing structure of large data sets. In Astronomical Data
Analysis Software and Systems XIV ASP Conference
Series, page 51.
Vera, L., Campos, R., Herrera, G., and Romero, C. (2007).
Computer graphics applications in the education pro-
cess of people with learning difficulties. Computers &
Graphics, 31(4):649–658.
Vila, J., Beccue, B., and Anandikar, S. (2003). The gender
factor in virtual reality navigation and wayfinding. In
Proceedings of the 36
Annual Hawaii International
Conference on System Sciences, page 7.
Wegman, E. J. and Symanzik, J. (2002). Immersive projec-
tion technology for visual data mining. Computational
and graphical statistics, 11:163–188.
Wilkinson, L. (1999). The grammar of graphics. Springer-
Verlag New York, Inc, New York, NY, USA.
Wong, P. C. and Bergeron, R. D. (1997). 30 years of
multidimensional multivariate visualization. In Sci-
entific Visualization, Overviews, Methodologies, and
Techniques, pages 3–33, Washington, DC, USA. IEEE
Computer Society Press.
Yamaguchi, T., Chamaret, D., and Richard, P. (2011). Eval-
uation of 3d interaction techniques for google earth
exploration using nintendo wii devices. In 6
national Conference on Computer Graphics Theory
and Applications (GRAPP’11), pages 334–338, Vil-
amoura, Portugal.
Zeleznik, R. C., Forsberg, S. A., and Strauss, P. S. (1997).
Two pointer input for 3d interaction. In symposium on
Interactive 3D graphics, pages 115–120, New York,
Virtual Reality Techniques for 3D Data-Warehouse Exploration