Information Technology Reference
In-Depth Information
of this distillation process leading from data through information to knowledge are
interesting patterns
. These are introduced as non-randomproperties and relationships
that are valid, novel, useful, and ultimately understandable (Fayyad et al.
1996
; Miller
and Han
2009
).
It is this notion of data mining that builds the theoretical underpinning of the
notion of movement mining used here:
Definition
Movement mining
aims for conceptualizing and detecting non-random
properties and relationships in movement data that are valid, novel, useful, and ulti-
mately understandable.
Even though the definition of data mining implies large data sources, the core
elements of the definition refer to qualities rather than quantities. Instead of defin-
ing movement mining through particular techniques or methods such as artificial
intelligence, machine learning, statistics, or database systems, this topic adheres to
a conceptual view of qualifying the outcomes of the analytical process. The move-
ment mining process aims for the ideal of finding properties and relationships, in
a wider sense, any form of structure in the data, patterns or trends, segmentations,
similarities, or clusters, that measure up to the given qualities.
The qualities
valid
,
novel
,
useful
, and
ultimately understandable
, in accordance
to Fayyad et al. (
1996
) and Miller and Han (
2009
, p. 3), are in the following illus-
trated for the special case of movement mining using the example of the movement
pattern leadership (
P5
. Andersson et al.
2008
).
Leadership
here is defined as the
situation when in a group of moving entities “one object is leading others”, in the
sense that this object spatially leads the way and the others follow for some time
(Fig.
3.2
).
•
valid
—properties and relationships should be general enough to apply to new
data, hence they should not just capture an anomaly or a peculiarity of the
current data. Although initially inspired by coordination in gray wolves
(Peterson et al.
2002
) or grazing heifers (Dumont et al.
2005
), patterns describ-
ing collective motion pattern such as
leadership
or
flock
(Laube et al.
2005
;
r
e
3
e
3
e
1
ϕ
e
2
e
2
e
4
e
4
e
2
e
4
e
1
e
1
e
3
t
1
t
2
t
3
e
4
follows nobody
e
4
follows nobody
e
4
follows nobody
e
1
follows
e
4
e
3
follows
e
1
e
1
and
e
3
follow
e
4
e
2
follows
e
3
e
1
,
e
2
and
e
3
follows
e
4
e
2
follows
e
1
and
e
4
e
3
follows
e
1
and
e
4
Fig. 3.2
Movement pattern example
leadership
, Andersson et al. (
P5
.
2008
)
