Beruflich Dokumente
Kultur Dokumente
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/229017694
CITATIONS
READS
39
4 authors:
Antonio Gonzalez-Pardo
SEE PROFILE
SEE PROFILE
Estrella Pulido
David Camacho
SEE PROFILE
SEE PROFILE
General Terms
Algorithms, Experimentation
Keywords
Virtual Worlds, Hierarchical Clustering, Normalized Compression Distance
1.
INTRODUCTION
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice and the full citation on the first page. To copy otherwise, to
republish, to post on servers or to redistribute to lists, requires prior specific
permission and/or a fee.
SMVC10, October 29, 2010, Firenze, Italy.
Copyright 2010 ACM 978-1-4503-0175-6/10/10 ...$10.00.
Figure 1: Screenshots of the dierent stages of a lecture on Greek history. At a rst stage (a), the teacher
explains dierent concepts about Ancient Greece. At a second stage (b) students are collaboratively working
to build a classical monument related to the concepts explained at stage (a).
2.
N CD(x, y) =
(1)
A back-end service built on the WampServer web development environment and a SQLite database engine
designed to store technical and user guides, educational documentation such as courses guides, multimedia, group information (avatar proles, passwords,. . . ),
and a set of tools that allow the storage of dierent
logs containing VW data (position, eye-gaze, chat,. . . )
extracted from avatars interactions in the VW. This
back-end service provides an interface to automatically
map data logs to an adequate format for data analysis
and statistical software tools (i.e. it is possible to select dierent features extracted from the logs and map
them into .ar format for the Weka toolkit, or to be
processed by using Matlab).
4. A LECTURE ON HISTORY
A Web portal that provides access to public documentation (courses, technical and user guides,. . . ) and a
user/group management service that allows for three
dierent roles in the platform (administrator, group
responsible (i.e. educators) and regular users (i.e. students). This service allows administrators to manage
and control the regular users enrolled in a experiment
(i.e. course).
1
As it is described in Section 2, by using the OpenSim Simulator, the V-LeaF platform oers a new environment where
dierent kind of islands (or simply sims) can be built for
dierent goals. Currently V-LeaF has deployed several islands related to educational topics and a medicine island. VLeaF has been widely used in cooperation with High School
institutions to provide a controlled access to educational
spaces. Each high school has its own island where teachers can control any educational activity that they design
http://this.ii.uam.es/vleaf/en/index en.jsp
10
(a)
(b)
Figure 2: Representation of the eye-gazing data. Figure (a) shows how the avatar view is not the same as
the user view. Figure (b) shows the same scene but from a dierent perspective. The angle formed between
the avatar and user view, called eye-gazing is represented as a red arrow.
public channel to give a lecture and communicate with all
the students at the same time.
In this study, a lecture was given in a Greek amphitheatre where the teacher explained some basic concepts about
ancient history, ancient Greek art, monuments and temples.
During the rst stage of the lecture, all the students should
be paying attention to the teachers explanation. As Figure
1 shows, avatars are in the amphitheatre and some of them
are looking at the teacher. This stage is very important because in the second part of the class, students will work in
groups to build some historical buildings proposed by the
teacher.
5.
Finally, the eye-gaze is a calculated parameter that depends on two factors. On the one hand, it depends on the
distance from the avatar to the teacher. On the other hand,
it will depend on the user viewpoint or the user camera orientation. It is important to take into account that in virtual
worlds, there are two types of views. The rst one is the
avatar view which is what the avatar should see. And the
other view is the user view, or camera view, which is what
the real user is seeing on the screen. This last view is the
one used to compute the eye-gazing, because the purpose is
to determine when the real user is paying attention to the
11
camera
separation
(2)
EyeGaze = arccos(
)
|camera| |separation|
is the vector dened by the user view and
Where
camera
and
angle formed by
camera
distance, and it is represented
by a red arrow.
150
6.
100
EXPERIMENTAL RESULTS
50
0
0
Distance
10
15
Figure 3: Representation of the avatars angle of vision (y-axe) and their distance to the teacher (xaxe), before the lecturer requires their attention.
60
50
EyeGaze
40
30
20
10
0
0
6
8
Distance
10
12
14
12
0.908
0.696
0.949
Yst2
0.947
student1
Yst8
Yst5
0.852
teacher
student2
student8
0.819
0.950
0.941
0.925
0.801
0.913
Yst6
Yst7
Nst4
student4
Nst1
0.944
student7
0.919
0.737
Yst1
student3
Nst5
0.943
0.895
Yst4
0.797
student5
student6
Nst7
0.846
0.719
0.929
Nst6
S(T)=0.997067
Nst3
0.890
0.921
0.945
Yst3
Nst8
Nst2
0.906
0.890
S(T)=0.994739
7. CONCLUSIONS
13
means that the distance from the teacher to the avatar is not
relevant when the student is looking at the teacher. Text exchanged through the chat can be used to classify students.
In this case, the lecture needs to be analyzed in order to
determine which student is talking about the concepts explained by the teacher. A hierarchical clustering over the
similarity metric of the chat record, shows the groups of
students depending on the contents of their conversations.
Nevertheless, a deeper study on the integration of dierent types of data needs to be performed to determine the
basic rules that dene whether a student is paying attention
or not. These rules could be used to generate recommendations to the teacher about the students that are not paying
attention to his/her explanations and to help him/her taking the best decision about the next action.
8.
ACKNOWLEDGMENTS
9.
REFERENCES
14