Analysing Team Formations in Football with the Static Qualitative
Trajectory Calculus
Jasper Beernaerts
1
, Bernard De Baets
2
, Matthieu Lenoir
3
, Kristof De Mey
3
and Nico Van de Weghe
1
1
CartoGIS, Department of Geography, Ghent University, Krijgslaan 281 (S8), 9000 Ghent, Belgium
2
KERMIT, Department of Data Analysis and Mathematical Modelling, Ghent University,
Coupure links 653, 9000 Ghent, Belgium
3
Department of Movement and Sports Sciences, Ghent University, Watersportlaan 2, 9000 Ghent, Belgium
Keywords: Football, Team Formation Analysis, QTC, Sports Analytics.
Abstract: In this paper, we introduce the Static Qualitative Trajectory Calculus (QTC
S
), a qualitative spatiotemporal
method based on the Qualitative Trajectory Calculus (QTC), for team formation analysis in football. While
methods for team formation analysis are mostly quantitative, QTC
S
enables the comparison of team
formations by describing the relative positions between players in a qualitative manner, which is much more
related to the way players position themselves on the field. To illustrate the method, we present a series of
examples based on real football matches of a 2016-2017 European football competition. With QTC
S
, team
formations of both an entire team as well as a smaller group of players can be described. Analysis of these
formations can be done for multiple matches, thereby defining the playing style of a team, or at critical
moments during a game, such as set pieces.
1 INTRODUCTION
In this paper, we introduce a new method for
analysing team formations in football, based on the
Qualitative Trajectory Calculus (QTC; Van de
Weghe, Cohn, et al., 2005). We start by giving a brief
overview of established methods for analysing team
formations in popular team sports and football more
in particular. After that, we present the static QTC
(QTC
S
), an extension of the calculus introduced by
Van de Weghe et al. in 2005. After presenting the
novel methodology, we illustrate the application of
QTC
S
for analysing team formations in football by a
series of real football examples. In the fifth section,
we discuss the applicability of the method, its
drawbacks and opportunities, before ending with a
conclusion.
2 STATE OF THE ART
Thriving on technological advances in tracking
technology and the opening up of different sports
branches to data gathering, sports analytics has
become a booming business in recent years (DOrazio
and Leo, 2010). Dozens of parameters from players,
such as speed, heart beat rate, transpiration level,
position, acceleration, jump height, goals scored,
attempts, tackles, etc. are being monitored during
training and matches of different sports. Even data at
team level, called collective variables, such as
formation, pass statistics, average positions, number
of shots and others are being gathered (Rein and
Memmert, 2016). In this overview, we will focus on
collective variables and more specifically on the
analysis of spatial formations in team sports, which
we will refer to as team formation analysis in the
remainder of this paper. Since almost all team
formation analysis methods, regardless of the sport,
use positional data of the players, we start by giving
a brief but focused overview of the state of the art of
team formation analysis in popular sports, before
providing a broader overview of the domain for
football. For a more general overview of all different
sports analytics methods in football, we refer to the
works of Rein and Memmert, 2016 and Memmert et
al. (2017).
Beernaerts, J., Baets, B., Lenoir, M., Mey, K. and Weghe, N.
Analysing Team Formations in Football with the Static Qualitative Trajectory Calculus.
DOI: 10.5220/0006884500150022
In Proceedings of the 6th International Congress on Sport Sciences Research and Technology Support (icSPORTS 2018), pages 15-22
ISBN: 978-989-758-325-4
Copyright © 2018 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
15
2.1 Team Formation Analysis
In American football, Atmosukarto et al. (2013) did
efforts for the automatic recognition of offensive
team formations, which they defined as The spatial
configuration of a teams players before a play starts
(Atmosukarto et al., 2013, p. 1) Their method
automatically detects when one of five reference
offensive team formations is achieved during the
game. The big difference with football, however, is
that a football game is more fluent and dynamic, thus
team formations tend to change more during the
course of the game (Atmosukarto et al., 2013). Team
formation analysis in volleyball has been conducted
by Jäger and Shöllhorn (2012). Because of the distinct
separation of a volleyball game in separate rallies,
Jäger and Shöllhorn used the positions of the players
at the start and end of the rallies instead of the average
positions during the rallies. Furthermore, they divided
the players into attacking and defensive groups,
analysing the shape of the two groups separately. On
top of that, they discovered that, given a dataset with
formations of six teams, an unknown team formation
could be correctly classified/assigned to one of the six
teams. In basketball, Lucey et al. (2014) analysed
defensive team formations of basketball players in the
three seconds leading to a three-point shot attempt,
finding they were able to predict whether the team
was going to give up open shot opportunity or not.
At this point, we would like to stress the
difference between team formation analysis, which is
the topic of this paper and is a spatial type of analysis,
and the analysis to choose the optimal line-up of a
team, which can benefit from player-specific data.
The latter type of analysis focuses on the selection of
actual players for each of the positions on the field
and has been investigated more rigorously in, for
example, hockey (Colleen Stuart, 2017), football
(Barrick et al., 1998; Tierney et al., 2016), volleyball
(Boon and Sierksma, 2003), basketball (Dezman et
al., 2001) and cricket (Ahmed et al., 2013).
2.2 Team Formation Analysis in
Football
In football, teams generally aim to play according to
a specific team formation (Kaminka et al., 2003;
Kuhlmann et al., 2005), which can be defined as A
specific structure defining the distribution of players
based on their positions within the field of play
(Ayanegui-Santiago, 2009, p. 1). Advantages of one
specific team formation with respect to others, e.g.
increased running distances when playing against a 4-
2-3-1 instead of a 4-4-2, have been described by
Carling (2011). Mapping the advantages of different
team formations can be useful when comparing them
with the own team strengths and weaknesses in order
to choose the most suitable team formation for a
game. Team formation analysis in football can be
performed in various ways, based on different key
performance indicators that are derived from the
players positions (Memmert et al., 2017). For
example, Sampaio and Macãs (2012) suggested the
team centroid, team entropy, a team stretch index and
the surface area of the team as key performance
indicators for team formation analytics. Going further
on this, Frencken et al. (2012) added the inter-team
distance, i.e. the distance between the centroids of
both teams, as a key performance indicator to detect
goals or attempts in a match. Lemmink and Frencken
(2013) demonstrated the possibility to use these key
performance indicators not only for the entire team
but also for subsets of the team such as players with
specific roles, e.g. attackers or defenders.
A method for automatic detection of the type of
team formation based on the average position of the
players was proposed by Bialkowski et al. (2014).
They argue that, because of the players swapping
positions during the game, static ordering of the
players does not accurately represent the team
formation. In order to cope with this, they introduce
dynamic ordering of players by the role that they
occupy at a given instant in time. Using data from a
whole Premiere League season, Lucey et al. (2013)
and Biakowski et al. (2014) found no significant
difference between formations of different teams, but
could detect that English Premier League teams used
more offensive team formations during home games.
Various new methods use principles of (artificial)
neural networks (McCulloch and Walter, 1943).
Visser et al. (2001) used artificial neural network
systems to recognize the team formation of the
opponent team. Starting with the positions of the
opponent players at a certain timestamp, the neural
network tried to classify that moment into a set of
predefined team formations (Atmosukarto et al.,
2013) later used an analogue method in American
football) and proposed the appropriate counter team
formation for the own team. Going further on this
work, Ayanegui-Santiago (2009) proposed to include
multiple relations between players for the recognition
of team formations. He divided the players into three
groups (defenders, midfielders and attackers) and
used labelled graphs between nodes of adjacent
groups to describe and compare team formations.
The methods mentioned above generally aim at
calculating frequencies of team formations. This
facilitates comparison of different team formations
icSPORTS 2018 - 6th International Congress on Sport Sciences Research and Technology Support
16
and the temporal evolution of these team formations
during the game (Grunz et al., 2012). Furthermore,
the occurrence of team formations can be linked to
scoring goals and winning games, thus measuring the
success of a specific team formation for a team.
However, while most methods use quantitative
metrics, Perin et al. (2013) argue that quantitative
analysis is not sufficient to understand the team
formation of a game or a whole season.
Unfortunately, qualitative team formation analysis in
football is currently mostly performed by human
experts and is thus very labour intensive (Bialkowski
et al., 2014). The goal of this paper is to contribute to
this domain, by introducing QTC
S
for (automatic)
team formations analysis in football.
3 METHODOLOGY
In this section, we introduce the novel methodology
for sports team formation analytics. We start by
giving a brief overview of QTC, followed by the new
variant (QTC
S
) that was created for this type of
research. Following, we present a series of possible
applications for the method in team formation
analysis in football.
3.1 The Qualitative Trajectory
Calculus
QTC is a qualitative calculus for describing
spatiotemporal relations between two or more
Moving Point Objects (MPOs). The most basic
variant of the calculus, QTC
B
, describes the
movement of a pair of MPOs during a time interval
by means of two QTC-characters (Van de Weghe,
Cohn, et al., 2005). Afterwards, multiple variants of
QTC were introduced, each named by adding the
initial(s) of the variants name to the abbreviation
QTC in subscript (Bogaert et al., 2007; Mavridis et
al., 2015).
3.2 QTC
S
While QTC typically describes movement between
multiple objects, it can be extended easily to a new
variant named QTC
S
(Static Qualitative Trajectory
Calculus), which describes static formations of point
objects (POs), which are players in our case. When
describing the formation of POs with QTC
S
, the lack
of movement is dealt with by constructing all possible
vectors between the POs (Figure 1a). Subsequently,
QTC
S
-relations between each pair of vectors are
Figure 1: A formation of four players (POs) on a football field at t
1
and the vectors between them (a). The construction of the
QTC
S
-relations between two vectors d and l, consisting of the QTC
S
-relation of vector d with respect to the starting point of
vector l and of the QTC
S
-relation of vector l with respect to the starting point of vector d. If the vector moves away from the
starting point of the other vector, the QTC
S
-relation is denoted by +, if the movement is towards it, the QTC
S
-relation is
denoted by -. If the movement is neither away nor towards the marker (thus perpendicular to the connecting line between
the two starting points), the QTC
S
-relation is denoted by 0 (b). The QTC
S
-matrix describing the full formation of the four
players, including all relations between all of the vectors (c).
Analysing Team Formations in Football with the Static Qualitative Trajectory Calculus
17
constructed similar to QTC
B
(Van de Weghe, Cohn,
et al., 2005), shown in Figure 1b for the vectors d and
l. The different QTC
S
-relations are stored in a QTC
S
-
matrix, where the first character in each cell is the
QTC
S
-relation of the vector in the row header with
respect to the vector in the column header, the second
character is the QTC-relation of the marker in the
column header with respect to the marker in the row
header (Figure 1c).
3.3 QTC
S
for Team Formation Analysis
By constructing a QTC
S
-matrix at different
timestamps, QTC
S
can be used to describe the team
formation at different moments in time. If the number
of players in the formation is identical at each of those
timestamps, the QTC
S
-matrices will have the same
dimensions and can be compared by calculating the
distance between them. The distance between two
QTC
S
-matrices is calculated by summing up the
pairwise distances between all of its elements (QTC
S
-
relations), thereby using the conceptual distance
between QTC-relations (Van de Weghe and De
Maeyer, 2005). By dividing the total distance
between two QTC
S
-matrices by the maximal possible
distance (depending on the matrix dimensions), the
relative distance is calculated. For easier
understanding, the relative distance is recalculated to
a similarity value between 0 and 1. The current
implementation of the methodology was done in the
Python programming language.
4 APPLICATIONS OF QTC
S
FOR
TEAM FORMATION ANALYSIS
IN FOOTBALL
In this section, we present a series of examples of the
QTC
S
-methodology for team formation analysis in
football. Considering the novelty of the method and
the lack of a good ground truth (Feuerhake, 2016), the
focus in this section primarily lies on introducing
rather than validating the results. All examples are
based on real football matches of a 2016-2017
European football competition, but are presented
anonymously for privacy reasons.
4.1 Full Team Formation
Often, trainers aim to use one or more predefined
team formation(s) for their field players (excluding
the goal keeper) according to the situation in the game
and the team formation of their opponent. By using
QTC
S
to describe both the desired team formation(s)
as well as the actual performed formation, an
evaluation of the team performance can be made.
Figure 2, for example, shows compliance (similarity)
of an anonymous team with a 4-4-2 team formation
during six different matches, analysed with a
temporal resolution of five minutes. The higher the
similarity in the graph, the more the actual team
formation resembled the theoretical 4-4-2 shown on
the right side, during the game.
Figure 2: Similarity of an anonymous team with a theoretical 4-4-2 formation during 6 matches, with a temporal resolution
of 5 minutes.
icSPORTS 2018 - 6th International Congress on Sport Sciences Research and Technology Support
18
4.2 Analysis of a Teams Playing Style
While in Section 4.1 similarity with one reference
team formation is calculated, it is also possible to
calculate similarities with all of the generally
accepted reference team formations (such as those
used in popular football simulation games). An
example of this can be seen in Figure 3, where for two
matches of an anonymous team, frequencies of the
most similar reference formation at every second of
the game are displayed, illustrating the variety of
team formations performed by one team during a
game or even between different games. As such, a
teams playing style, i.e. a set of regular played team
formations by a team, can be defined and compared
between teams and matches.
Figure 3: Frequencies of team formations played by an
anonymous team during two matches, ordered according to
the frequencies of match 1.
4.3 Parts of a Team Formation
Going more in detail, it can be interesting to analyse
how different groups of players of a team, e.g.
defenders and midfielders, each stick to their
theoretical formation during a match. Figure 4
displays the compliance of the midfielders and
defenders of an anonymous team with their respective
reference formation throughout one match, with a
temporal resolution of 5 minutes. It can be seen that
the defence much more sticks to its reference
formation throughout the game than the midfield,
which naturally has a more flexible and interchanging
character (Gonçalves et al., 2014). Between minutes
15 and 30 of the match, however, the only period
during which the displayed team conceded (multiple)
goals, both defenders as well as midfielders had the
highest deformations with respect to their reference
formations.
4.4 Analysis of a Team Formation at
Set Pieces
Team formations at set pieces, i.e. corners and free
kicks, are one of the most studied and trained aspects
of team formation in football (Sarmento et al., 2014).
By transforming both the desired as well as the actual
performed formations at set pieces into QTC
S
-
matrices, coaches can get an overview of whether and
to what extent the ideal trained-on formation of their
own team was achieved in real matches or get insight
into the tactics and regularly performed team
formations at set pieces of opponent teams.
5 DISCUSSION
In this paper, we presented QTC
S
, a qualitative
calculus that can be used for team formation analysis
in football. This method can easily be applied to other
sports and incorporates both inter-player coordination
as well as inter-team coordination (Memmert et al,
Figure 4: The similarity of midfielders and defenders of an anonymous team with their reference formation throughout one
game.
Analysing Team Formations in Football with the Static Qualitative Trajectory Calculus
19
2017). While it has some similarities with already
established methods for team formation analysis (see
Section 2), we are convinced of its added value by its
qualitative character, simplicity and extensibility.
With respect to the qualitative character, we feel
that quantitative methods fail to incorporate the
perception of the players positioning themselves into
the team formation on the field. These perceptions
will more likely be qualitative, e.g. I am too far
behind the opponents midfielder or I am standing
too close to my keeper than quantitative, e.g. I am
currently 21.45 meters away from my teams left
winger. As such we are convinced that qualitative
methods will better grasp the principles players use to
position themselves on the field. Although Ayanegui-
Santiago (2009) already proposed a similar
qualitative method, important differences in this
respect can be noticed. First of all, no distinction
between the studied players is made with QTC
S
,
drawing and comparing vectors between all the
players and thus using all spatial information for the
analysis. Secondly, QTC
S
results are standard
rotation-invariant, although rotation-sensitivity can
be enforced by adding static points (such as the
corners of the football field) to the QTC
S
description
of a team formation. Thirdly, we believe the QTC
S
-
methodology can calculate distance between different
formations more precisely, through the use of
conceptual distances (Van de Weghe and De Maeyer,
2005) between QTC
S
-characters (instead of the
duality between identity and non-identity between
characters) and the option to extend the number of
QTC
S
-characters used, conform the extension of
QTC
B
to QTC
B2
and QTC
C
(Van de Weghe, De Tré et
al., 2005). Moreover, Ayanegui-Santiago argues that
his work could be enhanced by the conversion of the
numerical orientations between players into
symbolical ones, such as the QTC
S
-characters.
In Section 4 of this paper, we presented a series of
applications of the QTC
S
-methodology in football.
The applications, however, are not limited to this list,
as one could for example analyse how a team gets
back into formation in the minutes after conceding a
goal or analyse how substitutions affect the quality of
the team formation, and so on. Furthermore, by
linking the team formation with performance factors
such as scored goals, won matches or ball possession,
coaches could be supported into making better
decisions and ultimately, try to win more games.
We are, however, aware of the lack of concrete
validation of the results in this paper, but would like
to point out the difficulty of validating methods for
(team) formation pattern detection in football due to
the lack of a good ground truth, as argued by
Feuerhake (2016). Furthermore, because of a huge
variety in methodologies, it cannot be assumed that
finding the same results as other established methods
is desirable nor that it should be the goal. As such, we
are convinced that validation should primarily be
done by sports professionals (e.g. coaches).
At the moment, the proposed methodology has
some computational limitations. Permutations
between players, for example, allowing to detect
similarities between formations where two or more
players switch roles, are possible though require high
processing power. As such, when using permutations,
the number of players that can be analysed is limited.
Furthermore, at the moment it is only possible to
compare formations with the same number of players
involved.
6 CONCLUSION
In this paper, we presented QTC
S
, a novel method for
team formation analysis in football. We explained the
principles of QTC
S
and illustrated its applicability by
a series of basic football examples. With QTC
S
, team
formations of both an entire football team as well as
a smaller group of players can be described. Analysis
of these formations can be done for multiple matches,
thereby defining the playing style of a team, or at
critical moments during a game, such as set pieces.
Further research could analyse the impact of static
points or the extension of QTC
S
to include more
characters on the team formation detection accuracy.
Furthermore, different weights could be allocated to
the vectors between players, according to their
importance on the field.
ACKNOWLEDGEMENTS
The authors would like to thank the Research
Foundation Flanders for funding the research of
Jasper Beernaerts.
REFERENCES
Ahmed, F., Deb, K., Jindal, A. (2013). Multi-objective
optimization and decision making approaches to cricket
team Selection. Applied Soft Computing, 13(1), 402-
414.
Ayanegui-Santiago, H. (2009). Recognizing Team
Formations in Multiagent Systems: Applications in
Robotic Soccer. In Nguyen, N.T., Kowalczyk, R.,
Chen, S.M. (Eds.). Computational Collective
icSPORTS 2018 - 6th International Congress on Sport Sciences Research and Technology Support
20
Intelligence. Semantic Web, Social Networks and
Multiagent Systems. Lecture notes in computer
science, 5796, 163-173.
Atmosukarto, I., Ghanem, B., Ahuja, S., Muthuswamy, K.,
Ahuja, N. (2013). Automatic recognition of offensive
team formation in American Football plays. 2013 IEEE
Conference on computer vision and pattern
recognition, 991-998.
Barrick, M.R., Stewart, G.L., Neubert, M.J. (1998). Mount
relating member ability and personality to work-team
processes and team effectiveness. Journal of Applied
Psychology, 83, 377-391.
Bialkowski, A., Lucey, P., Carr, P., Yue, Y., Matthews, I.
(2014). “Win at Home and Draw Away”: Automatic
Formation Analysis Highlighting the Differences in
Home and Away Team Behaviors. Disney Research
Boon, B.H., and Sierksma, G. (2003). Team formation:
Matching quality supply and quality demand. European
Journal of Operational Research, 148(2), 277-292.
Bogaert, P., Van de Weghe, N., Cohn., A.G., Witlox, F., De
Maeyer, P. (2007). The Qualitative Trajectory Calculus
on Networks. In Barkowsky, T., Knauff, M., Ligozat,
G., Montello, D. (Eds). Spatial cognition V reasoning,
action, interaction. Lecture notes in computer science,
4387, 20-38. Springer, Berlin, Heidelberg.
Carling, C. (2011). Influence of opposition team formation
on physical and skill-related performance in a
professional soccer team. European Journal of Sport
Science, 11(3), 155-164.
Colleen Stuart, H. (2017). Structural disruption,
experimentation and performance in professional
hockey teams: A network perspective on member
change. Organization Science, 28(2), 283-300.
Dezman, B., Trinic, S., Dizdar, D. (2001). Expert model of
decision making system for efficient orientation of
basketball players to positions and roles in the game -
empirical verification. Collegium Antropologicum,
25(1), 141-152.
DOrazio, T., and Leo, M. (2010). A review of vision-based
systems for soccer video analysis. Pattern Recognition,
43(8), 2911-2926.
Feuerhake, U. (2016). Recognition of Repetitive movement
patterns-the case of football analysis. International
Journal of Geo-Information, 5(11), 208.
Frencken, W., de Poel, H., Visscher C., Lemmink, K.
(2012). Variability of inter-team distances associated
with match events in elite standard-soccer. Journal of
Sport Science, 30(12), 1207-1213.
Gonalves, B.V., Figueira, B.E., Mas, V., Sampaio, J.
(2014). Effect of player position on movement
behaviour, physical and physiological performances
during an 11-a-side football game. Journal of Sports
Sciences, 32(2), 191-199.
Grunz, A., Memmert, D., Perl, J. (2012). Tactical pattern
recognition in soccer games by means of special self-
organizing maps. Human Movement Science, 31(2),
334-343.
Jäger, M.J., and Schöllhorn, W.I. (2012). Identifying
individuality and variability in team tactics by means of
statistical shape analysis and multilayer perceptrons.
Human Movement Science, 31(2), 303-317.
Kaminka, G.A., Fidanboylu, M., Chang, A., Veloso, M.M.
(2003). Learning the sequential coordinated behavior of
teams from observations. In Kaminka, G.A., Lima,
P.U., Rojas, R. (Eds.). RoboCup 2002: Robotic soccer
world cup VI. Lecture notes in computer science, 111-
125. Springer, Berlin, Heidelberg.
Kuhlmann, G., Stone, P., Lallinger, J. (2005). The UT
Austin villa 2003 champion simulator coach: A
machine learning approach. In Nardi, D., Riedmiller,
M., Sammut, C., Santos-Victor, J. (Eds.). RoboCup
2004: Robot Soccer World Cup VIII. Lecture notes in
computer science, 3276, 636-644. Springer, Berlin,
Heidelberg.
Lemmink, K., and Frencken, W. (2013). Tactical
performance analysis in invasion games: Perspectives
from a dynamical system approach with examples from
soccer. In McGarry, T., O’Dono-ghue, P., Sampaio, J.
(Eds.). Routledge Handbook of Sports Performance
Analysis (pp. 89-100). London: Routledge.
Lucey, P., Bialkowski, A., Carr, P., Morgan, S., Matthews,
I., Sheikh, Y. (2013). Representing and discovering
adversarial team behaviors using player roles. 2013
IEEE Conference on computer vision and pattern
recognition, 2706-2713.
Lucey, P., Bialkowski, A., Carr, P., Yue, Y., Matthews, I.
(2014). “How to get an open shot”: Analyzing team
movement in basketball using tracking data. Disney
research.
Mavridis, N., Belloto, N., Iliopoulos, K., Van de Weghe, N.
(2015). QTC
3D
: extending the qualitative trajectory
calculus to three dimensions. Information Sciences,
322(1), 20-30.
McCulloch, W.S., and Pitts, W. (1943). A logical calculus
of ideas immanent in nervous activity. Bulletin of
Mathematical Biophysics, 5(4), 115-133.
Memmert, D., Lemmink, K.A.P.M., Sampaio, J. (2017).
Current Approaches to Tactical Performance Analysis
in Soccer. Sports Medicine, 47(1), 1-10.
Perin, C., Vuillemot, R., Fekete, J.D. (2013). SoccerStories:
A kick-off for visual soccer analysis. IEEE Trans. Vis.
Comput. Graph, 19(12), 2506-2515.
Rein, R., and Memmert, D. (2016). Big data and tactical
analysis in elite soccer: future challenges and
opportunities for sports science. Springerplus, 5(1),
1410.
Sampaio, J., and Macãs, V. (2012). Measuring tactical
behaviour in football. International Journal of Sports
Medicine, 33(5), 395-401.
Sarmento, H., Pereira, A., Anguera, M., Campanico, J.,
Leitao, J. (2014). The coaching process in football A
qualitative perspective. Montenegrin Journal of Sports
Science and Medicine, 3(1), 9-16.
Tierney, P.J., Young, A., Clarke, N.D., Duncan, M.J.
(2016). Match play demands of 11 versus 11
professional football using global positioning, system
tracking: Variations across common playing
formations. Human Movement Science, 49(1), 1-8.
Analysing Team Formations in Football with the Static Qualitative Trajectory Calculus
21
Van de Weghe, N., Cohn. A.G., De Maeyer, P., Witlox, F.
(2005). Representing moving objects in computer-
based expert systems: the overtake event example.
Expert Systems with Applications, 29(4), 977-983.
Van de Weghe, N., and De Maeyer, P. (2005). Conceptual
neighbourhood diagrams for representing moving
objects. Lecture Notes in Computer Science, 3770, 228-
238.
Van de Weghe N., De Tré, G., Kuijpers B., De Maeyer
P. (2005). The double-cross and the generalization
concept as a basis for representing and comparing
shapes of polylines. In Meersman, R., Tari, Z.,
Herrero, P. (Eds.). On the move to meaningful
internet systems 2005: OTM 2005 workshops.
Lecture notes in computer science, 3762, 1087-1096.
Springer, Berlin, Heidelberg.
Visser U., Drücker, C., Hübner, S., Schmidt, E., Weland,
H.G. (2001). Recognizing formations in opponent
teams. In Stone, P., Balch, T., Kraetzschmar, G.
(Eds.). RoboCup 2000: Robot Soccer World Cup IV.
Lecture notes in computer science. Springer, Berlin,
Heidelberg.
icSPORTS 2018 - 6th International Congress on Sport Sciences Research and Technology Support
22