ArticlePDF Available

Match & Player Analysis in Soccer: Computer Coding and Analytic Possibilities

Authors:

Abstract and Figures

In team sports, especially in soccer, there is a certain complexity in the interaction between variables (players), variables of execution (displacements, lacks that will give place to regulation interruptions, throw, ball losses, recoveries, passes, etc.), and contextual variables (area of reception of the ball, area of pass of the ball, duration of the actions, etc.).
Content may be subject to copyright.
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
M d
M. Teresa Anguera-Argilaga , Ángel Blanco-Villaseñor , José Luis Losada-López1, Toni
Ardá-Suárez2, Oleguer Ca 4, Antonio Hernández-
1University of Barcelona, Barcelona, Spain, 2INEF-Galicia, A Coruña, Spain,3INEF-Centre
a, Spain,4IVEF, Vitoria, Spain,5University of Málaga, Málaga, Spain,6Iceland
University, Reykjavík, Iceland
ss area, displacement,
interruption, interception, shot, duration of the action, and marker. Each one of them, to the
eption, gives place to a system of categories that, logically, will be exhaustive and
mutually excluding. Also, it will record type of competition, pitch position, and the number
atch & Player Analysis in Soccer: Computer Coding an
Analytic Possibilities
1 1
merino-Foguet3, Julen Castellano-Paulis
Mendo5 & Gudberg Jonsson6
de Lleida, Lleid
In team sports, especially in soccer, there is a certain complexity in the interaction between
variables (players), variables of execution (displacements, lacks that will give place to
regulation interruptions, throw, ball losses, recoveries, passes, etc.), and contextual variables
(area of reception of the ball, area of pass of the ball, duration of the actions, etc.).
In order to achieve a systematized game analysis, we propose an instrument (SOF-3) that
combines a structure of field formats with systems of categories, and it’s substantially
different that SOF-1 (Anguera, Blanco, Losada, Ardá, Camerino, Castellano y Hernández
Mendo, in press; Anguera & Jonsson, 2002). The approaches of the instrument are: Times of
the party, player, lateral spaces, area of reception of the ball, pa
player's exc
of order of the plays of each one of the two teams, with the purpose of having a great
database coming from different seasons (complete seasons) and from different teams. The
code will be binary (0/1), and it will be carried out by computer, through software Excel
(Figure 1), and, also, through software ThemeCoder (Figure 2).
Figure 1. Excel record with SOF-3 instrument.
118
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
119
About this instrument (SOF-3
Figure 2. ThemeCoder record with SOF-3 instrument.
), we have built a codifying manual, with some sintactic rules.
ormation resides in the nature of the data, and in that facilitates, by
rmation in the analysis situation.
udy we apply the loglineal modelization that describes symmetrical relations
elop to this technique procedure CATMOD
lysis of variance by maximum likelihood
ects including of the analyzed model. The
itnessness of adjustment, and the statistical
icant effects that they contribute to this
of the raised model (in our case triple
eters, or, alternatively, they calculate odds
model to know in each one them the degree
en the diverse variables.
els uses several methods to study the pattern
in the contingency tables 2x2 (Cohen, 1980,
the mosaic shown by Hartigan and Kleiner
alues of any loglineal model. In the mosaic
The wealth of obtained inf
means of univariant and multivariant analysis, the search of diverse relationships among the
suitable approaches of the instrument. The main techniques of analysis of data that will be
used, will depend on the function of the characteristics of each one of the variable
(approaches of the instrument), and they will facilitate the decisions taking, keeping in mind
all the available info
In this st
between the variables. It has been used to dev
(SAS Institute, 1989). The table of the ana
contributes information for each one of the eff
probability reason is an indicator of the goodf
Wald indicates the significant and nonsignif
adjustment. Once determined the adjustment
association), they have estimated their param
ratio of the different effects that participate in the
intensity that is pronounced in the relations betwe
The graphical representation of the loglineal mod
of the association between the rows and columns
Friendly, 1991, Snee, 1974). The technique of
(1981), can be used to represent the residual v
each cell of the table is represented by a rectangle or area, but each one of them is
proportional to the frequency of the cell. The mosaic is built dividing a square in a vertical
sense, representing a variable (Balloon Contact category), and it will be divided in an
horizontal sense representing the other variable (Interruption category).
Another alternative of graphical representation consists of circumferences that represent the
residual values fit based on the intensity (Leverage) of the effects of model (D value of
Cooks).
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
References
a, M.T., Blanco, A., Losada, J.L., Ardá, T., Camerino, O., Castellano, J. y HAnguer ernández
igital de Alto Rendimiento en Fútbol.
Anguera, M.T. & Jonsson, G.K. (2002, June). Detection of real-time patterns in sports:
Interactions in football. Third Meeting of the European Research Group on
“Methodology for the analysis of social interaction”. Milan: Catholic University of
Milan.
Cohen, A. (1980). On the graphical display of the significant components in a two-way
contingency table. Commun. Statist.-Theor. Metho. A9, 1025-1041.
Friendly, M. (1991). Mosaic displays for multi-way contingency tables. York Univ.: Dept. of
Psychology Reports, No.195.
Hartigan, J,A., Kleiner, B. (1981). Mosaics for contingency tables. En W.F. Eddy (Ed.),
Computer Sicience and Statistics: Proceedings of the 13 th Symposium on the
Interface. New York: Springer-Verlag
SAS Institute, Inc. (1989). SAS/STAT User Guide Versión 8.2 (Vol. 2). Cary, North Carolina:
SAS Institute Inc.
Snee, R.D. (1974). Graphical display of two.way contingency tables. The American
Statisticican,28, 9-12.
Mendo, A. (in press). Instrumento de codificación y registro de la acción de juego en
fútbol (SOF-1). Revista D
120
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
C
Facu
le was modeled with sinusoidal functions. These
that the body center of mass reached maximum jump height.
different heights with differe es from were simulated
experimental data (Ar lidate the model. Three
ut the effect of series elastici ance were tested. First if a
able amount of series elastic energy econd
rk a ity and third if the CE can
del ehavior as the drop jump
discuss th energy processes in the contractile
es elastic elements of the muscle. A gy 30 percent of the
ody positive jumping energy was stored in lements (SEE). The
elocity of the CE in concentric was lower t scle tendon unit so that due to
ed. With the help of the SEE some
muscles are able to function at their optimal length, but this was not true for all muscles in the
model.
omputer Simulation of Muscle Elastic Behaviour in Drop
Jumping
Böhm, H.; Senner, V.
lty of Sport Science, Department of Sport Equipment and materials
Technische Universität München, Germany
It is well known that tendon and tendinous tissue of the muscles have elastic properties,
which can be determined using ultrasonography (Kubo 2000). From experiments on isolated
frog muscles (Kawakami 2000), it is known that shortening of muscle fibres is influenced by
the muscle series elasticity. However the exact role in different movements of daily life is
still unclear. The purpose of this study is to give insight of the role of muscle series elasticity
in drop jumping with the help of computer simulation. Vertical drop jumps from 20 cm are
simulated using a two dimensional forward dynamic model. The model consists of 4 rigid
bodies and 8 functional grouped one and two joint Hill-type muscles around hip, knee and
ankle joints. The active state of the musc
functions were optimized, so
Drop jumps from
and compared to
nt ground contact tim
ampatzis1999) to va
ty on jumping performhypotheses abo
consider can be stored and is released afterwards. S
if the contractile elements (CE) can wo
work in their optimal length. The mo
measured. Therefore it is valid to
t their optimal veloc
showed the same b
e mechanical
and seri considerable amount of ener
all muscles series elastic eb
v han that of the mu
the force velocity relation a higher force can be generat
Rectus
Iliopsoas
Gluteus
Gastrocnemius
Vasti
Tibialis anterior
Hamstrings
Soleus
Figure 1. Schematic representation of the musculoskeletal model used to simulate maximum-height drop
jumping.
121
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
References
Aram and
Kinesiology,11, 355-364, 2001.
Kawakami Y., Lieber R.L., J.Biomech. 33: 1249-1255, 2000.
Kubo K., Kanehisa H., Kawak Physiol. 81: 181-187, 2000.
patzis A., Schade F., Walsh M., Brüggemann G.P., J. of Electromyography
ami Y., Fukunaga T., Eur. J. Appl.
122
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
E-Learning Training for Physical Education Teachers:
The Experience of Veneto SSIS
Attilio Carraro1,2, Maurizio Bertollo2
1epartme eto, Italy
iod of stasis, in the last years a general renewal process in every level of the
primary school to university, has begun in Italy. Many
characterised in three training fields: a common field for
any line of study, with pedagogical, psychological lectures, and laboratories; specific fields of
each discipline, that includes historical and epistemological principles,
methodological studies and educational laboratories; training activities and professional
D nt of Educational Sciences University of Padua, 2SSIS of Ven
Introduction
After a long per
educational institutions, from
universities are currently engaged in planning and realizing specific postgraduate courses for
the initial preparation of teachers. Veneto SSIS (School of Specialization for Secondary
School Teachers) trains teachers for different subjects using two types of courses: traditional
in-presence courses and on-line courses in a blended modality. In this paper we present a
comparative study of these two teaching/learning methodologies. As regards teaching in-
presence, the on-line preparation allows an uninterrupted virtual contact between students and
professors and among students by means of e-forum and mailing lists. Through on-line
training, students can experiment constructive modality of cooperative learning that could be
more effective with respect to other traditional learning methodology.
ethodsM
In order to better influence improvement in the learning and teaching of all scholastic
subjects, including physical education, we think that careful attention should be paid to the
methodologies used for the initial preparation of teachers could be one of the major
challenge. Three values are particular relevant to the changes required for a quality physical
education teachers:
1. creating sufficient meaningful and challenging individual and collective learning
opportunities to improve professional practice;
2. supporting a sustained long term view of learning;
3. stimulating the cooperative and collaborative abilities.
The structure of The Veneto SSIS is
study for
practice. There is a close connection between theoretical teaching and professional practice.
The Veneto SSIS On-Line Project started in 1999 with eight courses that were fully agreed by
students; at present (2003) we offer 57 on-line courses for students of different subjects. In
the next year all the ten courses of the Physical Education area, excluding labs, will be
propose in the on-line format. These courses are organized in a blended modality (50% on-
line, 50% presence), including three moments strictly related: individual distance work,
interactive/cooperative distance work and the attendance of the traditional lessons. The
organization of an on-line course requires great resources (more than in the traditional
didactic) [see figure 1] and is characterized as a complex system of relationships between
students, professor and tutors, where every component interacts constantly with others and
where every action substantially modifies the behaviors of the others [see figure 2].
123
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
[Figure 1. The on-line organization]
SCIENTIFIC
CHIEF
EDUCATIONAL
MANAGER
ON-LINE
COURSES
COORDINATOR
WEB EDITOR
(information site)
WEB EDITOR
(courses
)
BSCW
MANAGER
INFORMATION
MANAGER
WEB
MASTER
ON-LINE
SECRETARY
STUDENTS
TUTORS
PROFESSORS
Tutors, on-line classes, web forum and news
is thesis.
(from 8/108 to 57/108), the total number of
less than 400 to more than
and s
student
our ye
ourses a questionnaire was administered to
assess the efficacy of the teaching/learning process. Some important considerations emerge:
100% of students participated in other on-line courses; 90% of students would advise it to a
friend; 60% of teachers declared who used interaction among colleagues more often than in
traditional courses. To conclude: compared to traditional teaching, e-learning promotes more
organized didactic work, a tendency towards a working habit in which the school is felt as a
knot of a net, a tendency teachers towards written report on their own school work and the
development of learning and training communities
groups allow a virtual/solid interactivity and
on-line project SSIS Community establishes
connection between university and schools,
between theoretical preparation and practical
training.
For a future physical education teacher,
conditio sine qua non is the ability to conduct,
facilitate relationships and begins productive
group processes. The central question is: can
the ability of educational mediation that is
greatly expressed in PE with a vicarious visual
and kinaesthetic formation also be taught
through a virtual formation?
Discussion
Observations, statistical data, final examination scores and post-impact analyses in the
schools suggest that e-learning can be an effective methods for initial and continuous
preparation of Physical Education teachers. In the next paragraph we present a few data to
support th
Results
In these last four years, we have recorded a
progressive growth of different indicators: the
proportion between total courses and on-line
accounts (from
11.000), the number of on-line tutors and,
unfortunately, the proportion between tutors
tudents. Through on-line training,
s can experimentstudent constructive modality
of cooperative learning that could be more
e with respect to other traditionaleffectiv
learning methodology. 132 Physical Education
/teachers have been engaged in the last
ars in on-line courses, at the end of allf
c
Text, documents
and other didactic
materials
Individual and group
exercises
FORUM
NEWS,
ORIENTATION,
TECHNICAL
SUPPORT E
RELATIONSHIP
and
DISCUSSION
Thematic
discussion
sub-groups
PROFESSOR and TUTOR
STUDEN
T
U
UN
NI
IV
VI
I
R
R
T
TU
UA
AL
L
/
/
S
SS
SI
I
S
S
12
4
individual
Working
Group
B
Bs
sc
cw
wB
Ba
as
si
ic
c
S
Su
up
pp
po
or
rt
tf
fo
or
r
C
Co
oo
op
pe
er
ra
at
ti
iv
ve
e
W
Wo
or
rk
kOn-line
classes
[Figure 2: On-line as a complex system]
124
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
R
Ca a
Rosenholt S. (1991) ‘Teachers’ W ization of Schools’, New York:
Teachers College Press.
eferences
rraro A., Zocca E., Lanza M., Bertollo M. (in press) ‘Problèmes épistémologiques de l
formation des professeurs d’éducation physique en Italie’, Paris: Ed. Revue.
orkplace: The Social Organ
125
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
An Information System for the Sport Scientific Theory of
Eder, C.; Strubreither, O.; Kornfeind, P.; Baca, A.
h Program "New Media in Education" by the Austrian Federal
Selected Sport Disciplines
Section of Biomechanics, Kinesiology and Applied Computer Science
Institut für Sportwissenschaft der Universität Wien, Austria
An information system is developed to transmit scientific basics of four selected sport
disciplines using internet and multimedia technology. The project is funded within the
framework of the Researc
Ministry for Education, Science and Culture. The aim of the project ("SpInSY") is to arouse
the interest in studying theoretical sport scientific concepts more intensively and to obtain a
more rapid understanding for complex connections. We expect to improve the
interdisciplinary understanding in sport science both in research and education and to achieve
a more economic and effective education.
Four modules are developed.
Sport disciplines Sub fields
Alpine Skiing
Tennis
Track and Filed Athletics (in particular
running and jumping disciplines)
occer
Sports Biomechanics
Sports Medicine and Exercise Physiology
Sports Psychology
Kinesiology
SApplied Computer Science
Exercise Science
Table 1: Modules of SpInSy and sport scientific disciplines covered
The online availability of the modules shall enable users to earn, refresh or deepen their
knowledge, independent of time and place. Computer animations, video sequences (e.g. to
control the ability to diagnose anomalies in motion performance or to specify the cognitive
representation of movements) and simulations (e.g. to illustrate the influence of different
parameters such as that of the effect of the ball rotation to the trajectory in tennis) are
included into the modules in order to achieve a high degree of interactivity.
he first module (Alpine skiing) is available since SeT
p
ptember 2002, the other modules are
articular finished. First evaluation results comply with the expectations.
126
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
A Method for Detecting the Impact Position in Table
Tennis
Kornfeind, P.; Baca, A.; Tutz, M.
Section of Biomechanics, Kinesiology and Applied Computer Science
Institut für Sportwissenschaft der Universität Wien, Austria
In order to identify strengths and weaknesses in the technical and tactical behaviour of table
tennis players process oriented models of the match are constructed. Using these models, the
temporal evolution of the match may be described. Parameters for describing the match
include the positions of the ball when hitting the table. Feedback systems giving the player
immediate acoustic or optical feedback in training on the position and/or quality of the ball
just played may be applied in training. Methods for detecting the position, where the ball hits
the table automatically (preferably in real time) would be advantageous in both cases. The
applicability of techniques based on vibration diagnosis has been investigated.
Three accelerometers (Type: Kistler 8632C10; four-channel amplifier 5134A1) have been
fixed onto the underside of one half of the table in the form of an isosceles triangle (one
corner near the middle of the net, two corners near the edges of the table). Vibration signals
board. All sig
vibration sign
instants of tim impact points were used to calibrate the acquisition
table and a comparison of
a real
me
produced by the ball hitting the table have been recorded using a NI-6062E data acquisition
nals were sampled at 160 kHz. A criterion for identifying the instant of time a
al arrives at a sensor has been developed. Repeated measurements of the three
e determined from known
system and to estimate the velocity of signal propagation. The accuracy of the system has
been tested by calculating the coordinates of another set of impact points from the three
instants of time measured.
The mean deviation (N=15) from impact points within the area defined by the position of the
sensors was 20.1 mm ( 9.7 mm). Impact points outside this area showed larger errors.
From this result it may be concluded that the method is accurate enough inside the area
spanned by the sensor positions. Additional tests to be performed will include the use of a
ourth accelerometer to extend this area to the total half of thef
calculated impact positions with those obtained by using a highspeed video system in
playing situation. Problems still to be solved are the calculation of the positions in real ti
and the perception / elimination of external influences, such as the stamping of table tennis
players when serving.
127
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
Virtual Reality: Visualization for Athletic Competition
R
University of Calgary, Calgary, Alberta, Canada T2N 1N4
EYWORDS: VIRTUAL REALITY, VISUALIZATION, ATHLETIC PERFORMANCE
have characteristics in common with
E’s created for entertainment and education? For example, virtual environments have been
the 2002 Olympics,
in the virtual
environment, how the simulated and the actual perience compared, how they felt the use of
petition, and what improvements could be
ichard M. Levy1, Larry Katz2, Ruth Morey Sorrentino2 and Xiufeng Peng1
1Faculty of Environmental Design, University of Calgary, 2Faculty of Kinesiology,
K
Introduction
This paper examines the user’s response to the representation of reality in a virtual world.
Specifically, this research considers if visual environments can be created with sufficient
realism to prepare an athlete for a first-time event in a foreign venue. Visualization
techniques have been a critical component in training athletes for the psychological stress
associated with a competition. Studies reveal that athletes’ performance can benefit from
programs that use some aspect of visualization (Hall, 1998; Martin, Moritz, & Hall, 1999). In
events where the difference between the top placing athletes is less than one tenth of a
second, preparation with visualization can be critical in winning an event. Evaluating the
effectiveness and potential benefits of using different simulation environments in athletic
training programs is a major focus of the Virtual Reality in Sport Research Group at the
University of Calgary.
A major goal of this research was to learn from interviews and open-ended questions, how
these virtual environments (VE) could be used in athletic training programs. More
specifically, do virtual environments created for sports
V
identified as having greater effectiveness in specific subject areas that involve the exploration
of a spatial or temporal dimension (Fuhrmann & Griemel, nd.). In creating versions of the
speed skating simulator, attempts were made to incorporate features found in action games
and simulators. Qualitative findings should help in defining a potentially useful environment
for training, while suggesting the best method of employing these virtual environments as
part of an overall training program.
Methods
Five Athletes preparing for the Olympic Speed Skating Competition in Salt Lake City were
given the opportunity to experience the oval space in a virtual environment several weeks
before the event (Figure 1). Written observations were made during the sessions held
between the sports psychologist and athlete. The researchers noted how the athlete felt at the
beginning of the session, how they chose to use the virtual environment, and their general
reaction to the virtual environment at the end of the session. In all cases, the athlete was free
decide if and how they used the virtual environment. On return fromto
athletes who used the virtual environment were asked if they would attend a de-briefing. At
these sessions, the participants were asked about the level of realism rendered
ex
the VE impacted their performance at the com
made to the VE.
128
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
Results
Of the five athletes participating in the experiment, four were seasoned competitors with five
or more years on the Canadian team. Overall, the participants felt the VE was realistic and
provided a reasonable likeness of the Salt Lake Olympic Oval. One issue raised by the study
is whether the quality and length of the sound tracks can impact the user’s response to the
VE. In creating a VE, sound loops are often utilized for efficiency. However, as one athlete
noted the sound loop can be repetitive and distracting. In addition, attention to the visual
details in the VE must be very accurate. As one skater remarked, the banner and flags in the
VE were not the same as those displayed at the Olympics in Salt Lake. Even details such as
having a skater moving in the wrong direction in the warm up lanes can create a
realism in the sense of motion of the virtual
speeds. Without a feeling of physical acceleration, it may be very difficult to
d for
heir race
in
that
preparing skaters in a virtual environm
anxiety levels before an actual long track speed skating event.
References
Fuhrmann, A. & Griemel, B. (n.d.) Teaching in Virtual Reality-Concepts and Evaluations,
VRVis Center for Virtual Reality and Visualization &Vienna University of
Technology, n.p.
Martin, K., Moritz, S., & Hall, C. (1999). Imagery use in sport: A literature review and
applied model. The Sport Psychologist, 13(3), 245-261.
Hall, C. (1998). Measuring imagery abilities and imagery use. In J. L. Duda (Ed.) Advances
in Sport and Exercise Psychology Measurement (Morgantown, WV: Fitness
Information Technology, Inc. 165-172.
disconcerting response. Also noted was a lack of
skater at high
overcome a perception of moving too slow at higher speeds.
Discussion/Conclusion
One of the most important benefits noted by the skaters was the use of the VE as an ai
visuali e from start to finish. zation. Many of skaters find it difficult to visualize the entire rac
Having a virtual environment made it possible for the athlete to stay focused on t
strategy.
Knowl ed invaluableedge gained from working with the athletes and staff has prov
developing this project. Future efforts will be spent on developing a virtual environment
can be used to simulate the time before the race and the actual start. It hoped that by
ent for the unexpected, it may be possible to reduce the
Figure 1. Virtual enviroment along the skater’s path.
129
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
Application of an Internet Based, Collaborative Software
years, many methods for the genuine informational coupling of training and
odern information and communication technology, these problems
viated. The generic term groupware describes software systems, which support
eographically distributed people engaged with a common task, and provide an
To enable the German top level beach volleyball
hes via the internet, the university of Augsburg
S v.4 (Video-Analysis-System) [LL02], which
nalysis of beach volleyball gam s based on a
real-time video and audio conferencing; text
or bandwidth or high latency; (c) application
se n two
ementation of the developed software system will start in co-
each volleyball teams, during their Olympic qualification in the
ity of Augsburg, 2002.
Environment in Training and Competition of Top Level
Beach Volleyball
Link, D.; Lames, M.
Universität Augsburg, Germany
During the last
competition were suggested and applied to improve performance in sports. In the majority of
cases, the systematic analysis of the own and the opponent’s technical and/or tactical skills,
and conclusions on the training process, requires the physical presence of coaches at the
competition location. In semi-professional sports, this “on the spot” service causes serious
financial, logistical, and organisational problems due to restricted resources of players and
federations. By using m
can be alle
groups of g
interface to a shared environment [Chu01].
teams to work collaboratively with their coac
developed a groupware system, labelled VIA
provides the following features: (a) video a
qualitative observation method [HL01]; (b)
messages for internet connections with po
viewing/ sharing by broadcasting analysis res
software instances; (d) specialized whiteboards and tactic charts to demonstrate player
positions and their dynamic interaction. These key functions allow supplying players with
relevant information all around the world. Taking advantage of a central reporting station,
many teams could easily be equipped with scientific services without generating travelling
e
ults, videos and mou movements betwee
expenses. The practical impl
operation with the German b
summer 2003.
References
[Chu01]Churchill, E. F.: Collaborative Virtual Environments. Digital Places and Spaces for
Interaction. Springer-Verlag Telos, 2001.
[HL01]Hansen, G.; Lames, M: Die Qualitative Spielbeobachtung - Eine
Beobachtungsvariante zur Trainings- und Wettkampfsteuerung im Spitzensport
(Qualitative game observation – An observation procedure to support practice and
competition in top level sports). Leistungssport, 31(1): 63-70, 2001.
[LL02]Link, D.; Lames, M.: Interaktive Trainings- und Wettkampfbetreuung durch
Datenverbund. BISp-Projekt Internet-Training (Interactive support of practice and
competition in data networks. Project Internet-Training), unpublished progress report,
Univers
130
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
Soccer and Data Mining
{Emil-Mihai.Muresan, Bart.DeMoor}@esat.kuleuven.ac.be
ed in a
Muresan Emil-Mihai, Bart De Moor
Katholieke Universiteit Leuven, Belgium
The purpose of our project is the full and automated analysis of soccer games. Starting from
the video recording of a soccer game, the 2D coordinates of the players on the field and the
3D coordinates of the ball during the whole game are extracted by image processing and
stored in a database. The interesting actions from the videos can be labelled for easy retrieval
and statistics. The coordinates database and the annotation list are then processed and mined
for relevant information.
All interesting actions in the game video recording are labeled using VideoCoach©, a video
annotation tool that we developed. All annotated actions with their time stamps are sav
list, making it very easy to retrieve the desired actions from the videos, make compilations of
selected actions, compute statistics and more.
The game is recorded with four static cameras from the four corners of the field. Image
processing algorithms are used to extract player and ball coordinates in each frame of the four
videos and the coordinates are saved in a database. The software that we developed for the
first part, PlayerTracker, makes use of image processing, geometrical transformations and
interpolation techniques to extract coordinates from images. In the initialization phase the
user has to specify a mapping between the video image and the real soccer field. The player is
first detected in the image, and then its real field position is computed through interpolation
based on the user-defined mapping (see figure 1).
. Image to field mapping and real position computationFigure 1
Once we have the coordinates database, the analysis potential is huge. Player or team
performance, referee, ball trajectory, are just some directions of interest for analysis. Our
application already derives some very interesting features. Speed and acceleration of players,
field coverage of players (figure 2.a,b), preferred running lines for a player (figure 2.c),
detection of strategies and team line-up (figure 2.d), quantification of team offensiveness and
defensiveness (figure 2.g), detection of counter attack danger situations (figure 2.e), primary
passing channels (figure 2.f), offside detection, referee positioning with respect to the game
(figure 2.h,i), ball trajectory analysis, and many more.
131
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
Figure 2. a. Player field coverage; b. Player position cluster; c. Preferred running lines; d. Team line-up; e.
ces, not yet
accepted by soccer organizations.
References
Masson, F. (1998). Football +: Comment gérer une équipe?. Manual from the Belgian Heizel
Trainer School, Brussels, Belgium.
Counter attack danger detection; f. Primary passing channels; g. Offensiveness and
defensiveness of teams; h. Histogram of referee distance to the ball; i. Histogram of referee
distance to the main diagonal.
The analysis application combines mathematical methods with observations made by coaches
along the way. The coach’s expertise is crucial in obtaining a good analysis of the game, but
features can be quantified and analyzed mathematically. An example is the theory stating that
when the centres of gravity of the convex hulls of the two teams (figure 2.e) switch position
along the sideline of the field, there is a danger for counter attack [Masson, 98]. Another
example is shown in figure 2.i. The rules of refereeing say that the referee should run close to
the main diagonal of the field (which is the case for this referee).
Player coordinates open huge possibilities of analysis for soccer games (can be easily
extended to any sport). In addition, the image processing technique to extract them is non-
invasive, as opposed to hardware solutions where players wear some emitting devi
132
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
A
1,2 Nader Rahnama, Ph mas Reilly, PhD, DSc
1Research Institute for Sport & Exercise Sciences, Liverpool John Moores University,
Physical Education Department, Isfahan University, Isfahan, Iran
indicate whether the
team was playing at home or away. The field designated as ‘period’ allowed the analyst to
ng actions occurring in six 15-minute periods of the game. The field called
ying actions with respect to 18 different
analyst to code one of 16 playing actions
he key events in soccer games. The field of ‘risk of
injury’ allowed the analyst to code each playing action with regard to the likelihood of it
causing injury and also when injuries result, the severity of the injury. When each field had
been completed, clicking on ‘Add Record’ enters the data into the main table. The data can be
viewed at any time by opening the table and using the other buttons to move forwards and
backward through all of the records entered. Queries were constructed to extract the data in
relevant ways. A query can be constructed which has fewer categories. For example, a query
could be set up which has a specific team, action and injury category
Compared to a hand notation system, which has been adopted previously in assessing injury-
risk in soccer by Rahnama et al. (2002a), the computerised system has several
Computerised Notation and Analysis System in Soccer
D; 1Adrian Lees, PhD; 1Tho
Liverpool, UK
2
Soccer is associated with a high rate of injury (Dvorak and Junge, 2000; Rahnama et al.,
2002b). For prevention of injury, risk factors should be identified and notation analysis is
helpful for identifying precursors to injury (Reilly, 1993). Notation analysis and associated
computerised systems can provide an accurate and objective evaluation of game
characteristics, including injury and associated risk. Hand notation systems represent the
simplest form of behaviour analysis and have been used for many years. Reilly and Thomas
(1976) employed a motion analysis system in order to estimate the work-rates of soccer
players. Since then, hand notation has been complemented by computerised methods of
analysis for the rapid processing of data. For example, McGarry and Franks (1995) used a
computer-based system for notational analysis to create models of the game and to identify
winning tactics. There has been no serious attempt to design a definitive notation analysis
system for assessing injury or injury risk in soccer. A new hand notation method for analysis
of injury and injury risk has recently been introduced by Rahnama et al. (2002a). This
method was found to take a long time to record and analyse data and led to the design of a
novel computerised system. The system designed can be modified to enable rapid collection
and analysis of data for a wide range of sports injury contexts. The aim of this report,
therefore is, to describe and explain the computerised notation and analysis system that is
designed for analysing injury and injury risk in soccer.
The computerised system designed in Microsoft Access and events were classified into the
following categories: team; period of play; zone of the pitch; playing actions and risk of
injury (Figure 1). The field referred to as ‘team’ allowed the analyst to
code the playi
‘zone’ of play’ allowed the analyst to code the pla
allowed thezones on the pitch. The field of ‘action’
occurring in the game, which covered t
133
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
Figure 1. Data entry form of the computerised system
advantages. Data are easy to update and correct, data are stored more securely on a computer,
the analysis of the data is more flexible and quicker, it is easier for someone else to follow the
procedure involved and the results generated from a query are more reliably produced. In
particular, the system is based on a commercial database program and so can be easily
constructed and individualised for a specific requirement. It is concluded that the
computerised notation analysis system has strengths relating to the ease of use and speed of
data handling and can be used for research into sports injuries. It can be recommended to
e areas of injury and injury risk in sports clubs, particularly
physiotherapists and other medical team members for recording, analysing and storing
formance, 8, 113-129.
Rahnama, N., Reilly, T. and Lees, A. (2002a). A critical incident analysis of injury risk
etitive soccer play. Physical Therapy in Sport, 3, 44-45.
Rahnama, N., Reilly, T. and Lees, A. (2002b). Injury risk of playing actions during
Reilly,
(eds.), Science and Football II. E. and F. N. Spon, London, pp 3-11.
Reilly, T. and Thomas, V. (1976). A motion analysis of work-rate in different positional
roles in professional football match-play. Journal of Human Movement Studies, 2, 87-
97.
researchers who work in th
information related to injury and injury risk in their players.
References
Dvorak, J. and Junge, A. (2000). Football injuries and physical symptoms. A review of
the literature. American Journal of Sports Medicine, 28, S3-9.
McGarry, T. and Franks, I.M. (1995). Modelling competitive squash performance from
quantitative analysis. Human Per
in comp
competitive soccer. British Journal of Sports Medicine, 36, 354-359.
T. (1993). Science and football: an introduction. In T. Reilly, J. Clarys, A. Stibbe
134
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
M
Introduction
puter development various new presentation techniques have been
g,
r these
d
ic algorithms (Goldberg 1989) help to optimize the trajectory in
less time compared to deterministic optimisation methods (simplex algorithm or gradient
een calculated and real trajectories and run times of world class
eifriz 2001). The scientific findings and
References
Goldberg, D.E. (1989). Genetic algorithms in search, optimization and machine learning.
Bonn.
Seifriz, F. (2001). Simulation im alpinen Skisport. Berlin.
odelling in Sports: From Mathematical Fundamentals to
Applied Use in Mass Media
Seifriz, F.; Mester, J.
Deutsche Sporthochschule Köln, Germany
Caused by modern com
made available in the last years. This refers to digital video as well as to modellin
simulation and animation of movements. Although the scientific background fo
techniques has been improved remarkably, there is a delay of penetration to mass media an
training in sport.
Methods
Biomechanical models and non-
conventional approaches by means of
genetic algorithms for determination and
simulation of performance were used to
calculate e.g. the optimal trajectory of a
skier. The necessary data for the individual
skiers result from anthropometrical
measurements, the slopes were surveyed with GPS. These data were processed in 3D-Studio
MAX and then used for visualization and animation of the runs (Fig. 1). Finally digital videos
were manipulated in terms of superimposing two runners into one video (Fig. 2).
Results and Discussion
Deterministic modelling the trajectory with individual parameters of the athletes and the
various conditions of the slopes faces the number, complexity and interdependence of
independent variables. Genet
Fig. 1. Virtual trajectory Fig. 2. Superimposing
method). A comparison betw
athletes demonstrate realistic settings (S
developments raised remarkable interest not only in the various groups of athletes but also in
mass media and supplying television companies. This adds new perspectives to research in
sport science as these techniques can contribute to an improvement of media presentation as
well as to an optimization of athletic training and performance. Especially non-conventional
approaches such as GA seem to promise new in-sights into the complex structure of sport
performance.
135
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
Handball Match Analysis: Computerized Notation System
Skarbalius, A.; Krušinskien , R.
Lithuanian Academy of Physical Education, Kaunas, Lithuania
Competition analysis provide information about past sport performances and serves for
predictive model development (Franks & McGarry 1996). Performance structure in handball
is of specifically complex character. One of the main solutions of these inherent problems has
been the use of notational computerized analysis systems, which allows generate and edit
match analysis reports directly after the match. Presently information on handball is most
frequently provided applying the WIGE MIC electronic version. Though this analysis of
matches is very wide and informative, it still lacks other extremely important factors
characterizing the performance of the team:
1. Match reports of attack actions in defence and offensive taken down in shorthand.
2. Factors characterizing the performance of the team: duration of attacks, ratio and
efficiency of positional and counterattacks, ratio and efficiency of counterattacks individual,
counterattacks group, counterattacks team, efficiency of actions in defence, features of
actions in the zone of 6-9 m.
The system worked out by us enables two experts to register match performance on offensive
and defense actions of every attack undertaken. User friendly and intuitive interface enables
to collect data during the match. Such data and its analysis have not been presented in
European, World and Olympic Games so far.
The program is developed as a standalone application. Data is stored in Microsoft Access
database files; program outputs are generated as Microsoft Word documents. Such data
organization allows quick generating and editing of match analysis reports, experts may
manipulate them and add comments or remarks. The system developed is mobile and adapted
for match environment.
KEYWORDS: HANDBALL, NOTATION SYSTEM, MATCH ANALYSIS, SPORT
PERFORMANCE, SOFTWARE..
References
Franks I. and McGarry T. (1996). The science of match analysis, in Science and Soccer, (eds
T. Reilly), E. & F.N. Spon, London, pp, 363-375.
136
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
Kinesiology in Swimming via Multimedia
Soons, B.; De Belder, C.; Colman, V.; Persyn, U.
Faculteit Lichamelijke Opvoeding en Kinesitherapie, K.U.Leuven, Belgium
To help an e preventing
injuries, a so ment. The
outputs of the research strategy, working hypotheses and results about propulsion concepts,
y characteristics (structure, flexibility and strength) have been implemented
xpert to optimize a swimmer’s technique and dry land training, while
-called expert system, based on kinesiological research, is in develop
balance and bod
via multimedia, including animations.
Since the seventies, from the analysis of + 15 Olympic finalists per stroke (Munich ’72)
working hypotheses could be formed and were tested by skilled swimmers experimenting
with various techniques and by modifying their body structure or flexibility artificially.
In the eighties, a quick video-digitizing system was developed specifically for the
symmetrical strokes. The system allowed to reconstruct the flexible body in water and air,
needed to define the style, and to calculate the velocity (V) of the body centre of mass
(CMbody) per phase, needed to define the effectiveness and to derive propulsion concepts.
In the same period, breaststrokers experimented with dolphin-like undulation, resulting in
more even V of the CMbody and thus in more effective styles. This influenced the rule
change (’87), allowing to launch the head below the water surface.
In the nineties, the techniques of 65 breaststrokers and 59 butterflyers at international level
were digitized and divided into style variants. In the extreme flat and undulating variants of
these strokes different propulsion concepts and balance mechanics could be confirmed.
Moreover, per breaststroke style different movement variables relevant for performance
ovements around champions in
e water surface (with bare feet and different fins), the application of various
could be specified. By using colours to visualise the water m
waving below th
working hypotheses related to undulation could be confirmed.
Many body characteristics of 574 subjects of at least national level were more relevant for
performance per style variant than per stroke. Based on an individual profile chart,
containing body scores, a performance calculation could be made per stroke and style variant
with a mean error of 2%. Even at international level still 15% of the swimmers do not use the
fastest style variant according to their body characteristics.
To familiarize with expert systems, after the preceding multimedia outputs, interactive case
studies of diagnosed swimmers of various performance levels were implemented.
137
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
Modelling of the Energy Metabolism in Cyclist Using
Ergometer Tests
Weber, S.; Platen, P.
Institute for Cardiology and Sport Medicine, Deutsche Sporthochschule Köln, Germany
In competitive cycling the available power and therefore available energy has a great
influence on the overall performance in competition. Therefore the diagnostic of the energy
metabolism in cycling sports has become a major field in theoretical and applied sport
science. With the common performance tests, it is impossible to evaluate the metabolic
background causing a specific physiological reply to a load applied to the system.
the model of the regulation and dynamics of muscular energy metabolism from
(Mader 1984), we made this model applicable for practical use. Lacking the calculation of the
dynamics of high-energy phosphate, it became possible to calculate the activation of
oxidative and glycolytic metabolism from easy measurable data. The needed input data are
maximum oxidative and glycolytic performance and the body mass. The maximum oxidative
and glycolytic performance was measured using two short high intensity tests on a cycle
ergometer. Body mass was recorded using a standard scale.
The modelling of the state of the energy metabolism is only possible up to power demands
equalling 2max. Presupposing that the anaerobic threshold is a state of the energy
metabolism where the production and oxidation rate of pyruvate are equal, it became possible
r output at this ed a non-linear
t AT. The transmission ratio of 2max changes to power
output at the AT was greater than the transmission ratio of maximum glycolytic power, which
n the transmission ratio of body mass (Weber 2003).
on of the presented model and ergometer tests build a tool to model the state of
Theref
Mader A. (1984). Habilitationsschrift. DSHS Köln.
Based on
OV
to calculate the powe metabolic state. All parameters show
influence on the power output a OV
was greater tha
The combinati
energy metabolism in steady state conditions for any cyclists undergoing the needed tests.
ore it is usable for precise performance analysis in lab and field.
References
Weber S. (2003). Dipl. A. DSHS Köln.
138
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
Intellectual Capital and Electronic Rights Management
Wigger, U.
Deutsche Sporthochschule Köln, Germany
Introduction
Members and students of any university produce a wealth of new research results, insights
and scientifically validated knowledge. Also, in the area of sports informatics, new models
and procedures for visual representations and large amounts of simulation models are
continuously being developed. Today, modern technologies offer ways to assure optimal use
and distribution of such intellectual assets.
Methods
The Information & Communication Technology Centre at the German Sport University
Cologne, in co-operation with IBM, is implementing two key technologies for intellectual
asset and rights management: the IBM Content Manager (CM) and the IBM Electronic Media
Management System (EMMS). The CM is an enterprise-scalable repository for virtually any
type of digital content, including HTML and XML web content, document images, electronic
office documents, and rich media such as digital audio and video. A single CM can support
multiple content stores distributed across the enterprise, or across the internet. The repository
provides check-in/check-out version control, object-level access control, and advanced
searching based on user-defined attributes. The IBM EMMS is a suite of enabling tools for
digital distribution of rich media content, including security, rights management, reporting,
and payment interfacing.
Results and Discussion
Business enterprises have made sizeable amounts of money by capturing their innovative
potential early, planning their patents, mining them for value, and mapping strategies for the
transformation of intellectual assets into capital (Davis&Harrison, 2001). The combination of
electronic content and rights management enables the creation of new business models, also
for enterprises such as universities. It allows for flexible digital rights management and helps
protect assets through their entire life cycle. The decision, however, which assets to share
widely and which to hold on to for patent protection or licensing lies within the institution
and/or the intellectual property owner.
References
Davis, J.L. and Harrison, S.S. (2001). Edison in the Boardroom: How Leading Companies
Realize Value from Their Intellectual Assets. John Wiley&Sons, Inc.. New York. NY.
139
International Journal of Computer Science in Sport – Volume 2/Edition 1 www.iacss.org
Determinin Using an
Angle-Driven Simulation Model
School of Sport and Exercise Sciences, Loughborough University, Loughborough, United
Kingdom
arameters may then be used in kinetically driven models.
gle time histories and their first two
derivatives were obtained by fitting the joint angle data, obtained from image analysis, with
lines. Simulated Annealing was subsequently used to minimize the difference
etween simulated and recorded performances of the running jumps in order to obtain spring
Results and Conclusion
The evaluation resulted in differences of 5% and 8% between simulation and performance of
the jumps for maximum height and distance respectively. Minimising the difference between
the actual and simulated performances for the jumps for height and distance separately
resulted in differences of 4% and 7% respectively. It was therefore concluded that the
method used was able to provide model parameters values which perform well in different
simulations.
g Subject-Specific Parameter Values
Wilson, C.; Yeadon, M.R.; King, M.A.
Introduction
Parameter values are required for computer simulation models of human movement. This
paper describes a procedure for determining personalised parameter values for elastic
elements using kinematically driven simulations to match performances of running jumps.
These p
Method
An eight segment, subject-specific, angle-driven computer simulation model of running
jumps was developed using the software package AutolevTM3. The eight segments
represented the foot, shank and thigh of the takeoff leg, the shank and thigh of the free leg,
and the trunk, upper arm and lower arm. Wobbling masses within the shank, thigh and trunk
segments of the takeoff leg were represented by mass-spring-damper systems. The ground-
foot interface was represented by two spring-damper systems situated at the heel and toe. In
order to make the model subject specific anthropometric measurements were taken on an elite
male high jumper and these were used to determine the inertia parameters used in the model.
High speed video and force data were collected during running jumps for both height and
distance performed by the same high jumper. Joint an
quintic sp
b
and damping constants for both the wobbling masses and the foot ground interface. A single
set of parameters were determined from one jump for height and one jump for distance using
a combined matching score.
140
... The most important of these are the inevitable multiplicity of the individuals involved, the diversity of existing interactive structures, the effect of context and interactive content, and both inter-and intrasession permanence and changeability, among many others. The study of social interaction is, thus, undoubtedly difficult from a methodological point of view, and in this regard, the contribution of observational methodology has proved decisive, whatever the situation (Anguera, 2003, 2005; Carletta et al., 2003). Many specific issues can be addressed through the study of social interaction, since the latter constitutes an extensive field from which varied elements may be selected. ...
Article
Full-text available
Traditional methods for quantifying sport performances are limited in their capacity to describe the complex interactions of events that occur within a performance over time. The following article outlines a new approach to the study of actions between players in team sports--mainly, soccer. Since the observational design is nomothetic, point, and multidimensional, an observational and data-collecting instrument has been developed. The instrument is mixed and combines a field format with a category system for game events, as well as an ad hoc instrument that considers the game actions of one or both teams, each recorded according to the same criteria. The article also outlines a new approach to the analysis of time-based event records--in this case, sports performance--known as T-pattern detection. The relevant elements of the T-pattern detection process are explained, and exemplar data from analyses of soccer matches are presented to highlight the potential of this form of data analysis. The results suggest that it is possible to identify new kinds of profiles for both individuals and teams on the basis of observational criteria and a further analysis of temporal behavioral patterns detected within the performances.
Article
Full-text available
The purpose of this study was to find patterns in the butterfly swimming technique, with an adaptation of the Behavioral Observation System Tech. This, as an instrument for ad-hoc qualitative analysis, enables the study of the stability of the technical implementation. When used in the training of swimmers, analysis can reduce the variability of behavioral tuning swimming technique. Through the analysis of temporal patterns (T-pattern) and a sequence of five cycles running at hand maximum speed, the behavior of four technical Portuguese elite swimmers, with a record of 259 alphanumeric codes and a total of 160 configurations, were studied. The structure of the original instrument, based on a mixed system of categories and formats Field, can record technical features, observed during the execution of hand cycles. The validity was ensured through the index of intra-observer reliability (95%) and inter-observer accuracy (96%). To detect patterns in each swimmer, the Theme 5.0 software was used, which allowed to identify the stable structures of technical performance within a critical interval of time (p <0.05) - t-patterns. The patterns were different, adjusting to the characteristics of technical implementation of the swimmers. It was found that the swimmer can create settings with different levels of structure complexity, depending on the implementation of changes within the hand cycle. Variations of codes in each configuration obtained using the SOCTM, allowed determining the differences between swimmers. However, the records showed a clear behavioral similarity when comparing the result with a general pattern of the butterfly technique. The potential quality of this instrument seems to be important due to the patterns obtained from a temporal sequence
Article
We report that a player's behavioral response to a preceding athletic event in competitive squash is consistent when competing against the same opponent (p > .25) but inconsistent when competing against different opponents (p < .25). Some evidence of systematic behavior within the inconsistent responses was identified, from which a consequent strategy was analyzed and tested using a stochastic (Markov) model. The a priori strategy, all things equal, would have been expected to enhance sport behavior in this instance. That sport data provide a mostly reliable signature of athletic response to a preceding behavioral event against the same opponent, but not against different opponents, bears practical relevance to current sport practice. Sport analysis can reliably assume a prescriptive application in preparing for future athletic competition, but only if consistent behavioral data can be established. The traditional planning of match strategies from a priori sport information (scouting) for later use against the same opponent would otherwise seem to be an expedient and necessary constraint.
  • Y Kawakami
  • R L Lieber
Kawakami Y., Lieber R.L., J.Biomech. 33: 1249-1255, 2000.
Die Qualitative Spielbeobachtung -Eine Beobachtungsvariante zur Trainings-und Wettkampfsteuerung im Spitzensport (Qualitative game observation -An observation procedure to support practice and competition in top level sports)
  • G Hansen
  • M Lames
Hansen, G.; Lames, M: Die Qualitative Spielbeobachtung -Eine Beobachtungsvariante zur Trainings-und Wettkampfsteuerung im Spitzensport (Qualitative game observation -An observation procedure to support practice and competition in top level sports). Leistungssport, 31(1): 63-70, 2001.
Interaktive Trainings-und Wettkampfbetreuung durch Datenverbund BISp-Projekt Internet-Training (Interactive support of practice and competition in data networks Football injuries and physical symptoms. A review of the literature
  • D Link
  • M J Lames
  • A Junge
Link, D.; Lames, M.: Interaktive Trainings-und Wettkampfbetreuung durch Datenverbund. BISp-Projekt Internet-Training (Interactive support of practice and competition in data networks. Project Internet-Training), unpublished progress report, Univers References Dvorak, J. and Junge, A. (2000). Football injuries and physical symptoms. A review of the literature. American Journal of Sports Medicine, 28, S3-9.
Science and football: an introduction Stibbe References Franks I. and McGarry T The science of match analysis
  • T Reilly
  • J Clarys
T. (1993). Science and football: an introduction. In T. Reilly, J. Clarys, A. Stibbe References Franks I. and McGarry T. (1996). The science of match analysis, in Science and Soccer, (eds T. Reilly), E. & F.N. Spon, London, pp, 363-375.
  • S Weber
Weber S. (2003). Dipl. A. DSHS Köln.
Edison in the Boardroom: How Leading Companies Realize Value from Their Intellectual Assets
  • J L Davis
  • S S Harrison
Davis, J.L. and Harrison, S.S. (2001). Edison in the Boardroom: How Leading Companies Realize Value from Their Intellectual Assets. John Wiley&Sons, Inc.. New York. NY.
  • Y Kawakami
  • R L Lieber
  • K Kubo
  • H Kanehisa
  • Kawak Physiol
Kawakami Y., Lieber R.L., J.Biomech. 33: 1249-1255, 2000. Kubo K., Kanehisa H., Kawak Physiol. 81: 181-187, 2000.
Simulation im alpinen Skisport
  • F Seifriz
Seifriz, F. (2001). Simulation im alpinen Skisport. Berlin. References Franks I. and McGarry T. (1996). The science of match analysis, in Science and Soccer, (eds T. Reilly), E. & F.N. Spon, London, pp, 363-375.