Database Reference
In-Depth Information
objects spend some minimum amount of time. The authors index trajec-
tories by velocity, then extend the Pointwise Dense Region (PDR) [62]
method to identify ROIs.
Similarly, the work by Giannotti et al. [22] combine aspects of mining
interesting locations, with understanding and predicting movement pat-
terns. The authors extend prior work on mining frequent patterns and
define a spatiotemporal sequence (ST-sequence), which is a sequence of
locations visited by a set of users with a given level of support. The goal
in this work, much like in the frequent itemset mining, is to identify a
frequently visited sequence of locations over time. The authors address
the problem when the locations, or regions-of-interest (ROI), are known
as well as when they must be extracted from the data. More recent work
has built upon these concepts of pattern mining in the spatiotemporal
domain as well [58, 72].
5. Discussion and Future Research Directions
Over the past two decades, we have seen significant advancements in
all areas related to managing spatiotemporal data. In this chapter, we
attempt to cover the state-of-the-art solutions to the current challenges
specific to managing spatiotemporal data. Our review focused on data
management techniques as these are fundamental to nearly all applica-
tions involving mobility data. Additionally, we also covered some of the
core and recent work on tracking mobile objects. Lastly, we introduced
some of the recent applications of mining spatiotemporal data to extract
interesting patterns.
Despite the multitude of work in these areas, new and challenging
problems are constantly being introduced. Below we briefly outline a
few potentially interesting areas of future research.
Combining spatiotemporal information with social net-
works:
Work on social network analysis in the recent years has
been plentiful due to the explosion of the availability of social data
from sites such as Facebook, Twitter, MySpace, and various other
relationship or communication networks. The analysis of users and
their connections has largely focused on the concept of homophily,
which is the tendency of individuals to connect to others that are
similar to themselves. However, physical space is another signifi-
cant factor which influences how users interact with one another.
Combining information about a user's movement with her social
network presents an exciting research direction [16, 96].
Data-driven Techniques:
Large quantities of spatiotemporal
data are becoming readily available through several research ef-
