Trajectories and Their Representations - Mobility Data - page 11

Both Chapter 3 (trajectory databases) Chapter 8 (visual analytics for tra-
jectories) of this topic follow the first approach, with two different methods.
They allow their users to analyze the trajectories in context, that is, by ana-
lyzing the relations between the moving object and the environment. On the
other hand, Chapters 6 and 7 on trajectory analysis by data mining follow the
second approach: The trajectories are first enriched and then mined. In contrast
to raw trajectory representations, we qualify as semantic any trajectory repre-
sentation that has been created by enriching and transforming a raw trajectory
in order to add more meaning. In this chapter, we focus on the second approach
and therefore explore which kinds of semantic trajectory representations are
needed.
To create semantic representations of trajectories, the trajectory management
system needs to have access to application contextual data, which typically
includes knowledge of geographical objects (i.e., objects that have a known
position in geographical space) and events in the region and time period tra-
versed by the trajectories. For example, in the tourists scenario we will naturally
assume that the knowledge about the city map is available so that, for example,
trajectory paths can be described in terms of streets and crossroads and the points
where people stop can be identified with places of interest, such as landmarks,
significant buildings, monuments, museums, shops, restaurants, cafes, and sports
centers. Information about ongoing events, for example, shows, fairs, concerts,
and football games, is to be collected too, as it may influence the organization of
tourist tours. We use the term contextual data repository to generically refer to
whatever external source that can be used by the application to enrich trajectory
data.
All the following kinds of information can be used together with raw trajec-
tories in order to get semantic representations:
The geo-objects representing the places of interest, roads, regions where the
trajectory passed;
The events related to the movement of the object;
The transportation means used by the person for moving;
The activities performed by the person or animal when (s)he stopped.
The geo-objects corresponding to the positions of a trajectory can be found by
a process called geo-localization , which is a usual technique for adding semantics
to spatial data. It maps spatial coordinates ( x, y ) to the corresponding geo-objects
of the contextual data repository. For instance, in Figure 1.1 , the coordinates of
the spatio-temporal positions where the tourist stopped for a while have been
mapped to the corresponding POIs: Eiffel tower, Palais Bourbon, Le Louvre
museum, and so on. The positions where (s)he was moving have also been
mapped to the corresponding street segments. Chapter 2 presents in detail this
geo-localization process. Similarly, it is possible to find which events correspond