Database Reference
In-Depth Information
2
Trajectory Collection and
Reconstruction
Gerasimos Marketos, Maria Luisa Damiani, Nikos Pelekis, Yannis
Theodoridis, and Zhixian Yan
2.1 Introduction
The research area of trajectory databases has addressed the need for represent-
ing movements of objects (i.e., trajectories) in databases in order to perform
ad hoc querying and analysis on them. During the last decade, there has been
a lot of research ranging from data models and query languages to implemen-
tation aspects, such as efficient indexing, query processing, and optimization
techniques.
This chapter covers aspects related to data collection and handling so as
to feed trajectory databases with appropriate data. We will also focus on the
step
trajectory reconstruction
of the
Geographic Privacy-aware KDD process
(illustrated in Figure
2.1
) emerged from the GeoPKDD project which proposed
some solid theoretical foundations at an appropriate level of abstraction to deal
with traces and trajectories of moving objects aiming at serving real world
applications. This process consists of a set of techniques and methodologies
that are applicable to mobility data and are organized in some well-defined and
individual steps that have a clear target: to extract user-consumable forms of
knowledge from large amounts of raw geographic data referenced in space and
in time. However, when mobility data are about individuals, data collection is
subject to privacy regulations and restrictions. To enable privacy-aware collec-
tion of position data, a complementary class of techniques is used, known as
location PETs
(privacy-enhancing technologies).
This KDD process can be applied to heterogeneous sources of mobility data.
The cellphone icon that is illustrated in Figure
2.1
could represent various data
sets coming from various devices. In Section
2.2
, we present such sources.
Before applying trajectory reconstruction techniques we may need to per-
form some basic trajectory preprocessing. This may include parameterized tra-
jectory compression (so as to discard unnecessary details and concurrently keep
23
