Movement Spaces and Movement Traces - Computational Movement Analysis - page 10

repeatedly produced making use of random walk algorithms and agent-based mod-
eling. The advantage of simulated synthetic movement data is that in contrast to real
observed movement data, the movement processes to be studied and the hence emerg-
ing movement data can be rigorously controlled, which builds a crucial precondition
for experimental evaluation of methods (see also Sect. 3.3 ) .
Table 2.1 finally illustrates that even though GPS-tracking still is the most frequent
way of tracking MPOs, alternative ways of tracking moving objects have emerged
through the technological advancement of location-aware mobile ICT devices.
2.2 Conceptual Models for Movement and Movement Spaces
The above overview illustrates that the modeling of movement means modeling
the moving entities, but equally important modeling the space they move in. The
varying characteristics of possible conceptual space models embedding a form of
movement—be it an animal habitat, a 3D building, or a complex urban transporta-
tion network—rules how the entities can move, and consequently impacts on the
computational analysis tools that are required and suitable for understanding that
movement. Hence, a critical modeling decision early in the CMA process is the
choice of the conceptual data model for the space embedding the movement under
study. 1
The review article ( P9 . Laube 2009 ) has proposed a categorization of six basic
conceptual movement spaces that are commonly found in CMA (see Fig. 2.1 ). Ani-
mals tagged with GPS receivers (e.g., migrating birds) move in an unconstrained
Euclidean space (a). Sometimes movement is constrained as non-swimming animals
will not enter a lake or in an indoor environment shoppers can't enter locked rooms
(b). Especially visualization applications favor the space-time cube metaphor follow-
ing Hägerstrand's Time Geography (Hägerstrand 1970 ) (c). Movement can also be
captured in discrete tessellations of space, for example, as a series of discrete steps
through a field representation of space (d). Location-aware mobile devices leave
digital traces as a sequence of visited GSM cells (e). Finally, human movement is
often tied to a transportation network, where movement can only occur along edges
between intersecting nodes (f).
This categorization proved to be useful for leading the crucial discussion about
conceptual data models in CMA, a traditionally data-driven research field often
accepting the data-inherent structures as unchangeable preliminaries and neglect-
ing the implications of conceptual design choices. As will be shown in the next
chapter about movement mining, the different conceptual movement spaces allow
for the detection of different movement patterns (Fig. 3.3 ) . The categorization
1 Note that this section is focused on how movement traces can be abstracted and represented in
spatial information systems. Other authors have put forward conceptual models for movement in
different contexts, such as, for example, for explaining organismal movement in movement ecology
(Nathan et al. 2008 ), discussed in the related work Sect. 2.4 .