Understanding Human Mobility Using Mobility Data Mining - Mobility Data - page 145

hasTime

Time

Move

Place

Stop

isIn

trajCompOfMove

fpFromStop
fpToStop

trajCompOfStop

Long Home Work

Trajectory Pattern

TrajBelongs

Frequent
Pattern

Systematic
Movement

HomeToWork
Behavior

Systematic
Behavior

Figure 7.5 A fragment of the ontology used in Athena to discover HomeToWork behavior

reasoning tasks. Athena represents application of domain knowledge into an
ontology and a mapping between ontology concepts to data and patterns is
defined. Ontology concepts represent the data (e.g., trajectories, roads), patterns
(e.g., flocks), and semantic behavior (e.g., StuckInTrafficJam). The ontology
embeds the semantic of the domain application and, particularly, the concepts
defined by axioms define the semantic behavior we want to infer from patterns
and data.
We clarify now using as an example the CommuterMovement behavior. We
can represent this behavior in the ontology as an axiom defining a trajectory
moving from outside the city in the morning, stopping a long time in the city
center, then moving back from center to the outside in the afternoon . The map-
ping between the trajectories, patterns, and the ontology is formalized in a
specific mapping file, so that the trajectories, the mobility patterns, and the geo-
graphical knowledge become instances in the ontology. The ontology reasoning
engine is run to classify patterns and trajectories into the appropriate behav-
ior as defined by the axioms (e.g., the trajectories satisfying the HomeToWork
behavior).
In Figure 7.5 we present an example of the ontology definition for the Home-
ToWork behavior. We can see that HomeToWork and SystematicMovement are
subclasses of trajectory because they represent individual behavior, while Sys-
tematicBehavior is defined as a special kind of frequent pattern, which in turn
is a kind of pattern, thus representing collective behavior.
A special function called SEMANTIC(object) is defined in M-Atlas with
the objective of returning all the ontology concepts in which the object has
been classified by the mapping file or by the inference engine. For example,
a given trajectory may belong to either the class “Trajectory” as defined by