Database Reference
In-Depth Information
Annotating trajectory with regions : This component enables annotation of
trajectories with meaningful geographic or application domain sources of
semantic regions. It does so by computing topological correlations between
trajectories and third party data sources containing geo-objects of regions
(called regions of interest or ROI). We need to design a spatial join algorithm,
which can work for both regular regions (e.g., 100 m × 100 m grid-based
land use data) and irregular regions (e.g., regions with free-style shapes such
as EPFL Rolex Learning Center).
Annotating trajectory with lines : This component annotates trajectories with
lines of interest (LOI) such as road networks and considers variations present
in heterogeneous trajectories (e.g., vehicles run on road networks, while
human trajectories use a combination of transport networks and walkways).
Given data sources of different forms of road networks, the purpose is to
identify correct road segments as well as infer the transportation modes such
as “walking,” “cycling,” and “public transportation” such as metro and bus.
Thus, the algorithms in this component include two major parts: the first part
is designing/reusing a global map-matching algorithm to identify the correct
road segments for the move episodes of a trajectory, and the second one is
inferring the transportation modes that the moving objects/people used during
their moves.
Annotating trajectory with points : This component annotates the Stop epi-
sodes in trajectory using information about suitable points of interest (POIs).
Examples of POI are “restaurants,” “bars,” “shops,” and “movie theaters.” For
scarcely populated landscapes, it is relatively trivial to identify the objective
of a stop (e.g., petrol pump on a highway, back home in a very sparse resi-
dential area). However, densely populated urban areas bring many different
types of candidate POIs for a trajectory stop. The problem of inferring stop
behaviors using POIs becomes challenging. Further, low GPS sampling rate
due to battery outage and GPS signal losses makes the problem more intri-
cate. Recently, a HMM (hidden Markov model)-based inference algorithm
has been designed to extract the underlying stop behaviors in the trajectory.
In this algorithm, the location of individual trajectory stop is modeled as a
model observation, whilst the POI category is considered as the hidden state
that needs to be extracted.
2.5 Protecting the Privacy of Individuals' Positions
This section overviews techniques that aim at protecting users' privacy during
the data collection process. The concern for privacy stems from the fact that
whenever position refers to individuals, position is qualified as personal data,
and collecting personal data is restricted by privacy norms and law in several
Search WWH ::




Custom Search