Database Reference
In-Depth Information
1. Introduction
We consider the problem of discovering similar time-series, especially under
the presence of noise. Time-series data come up in a variety of domains,
including stock market analysis, environmental data, telecommunication
data, medical and financial data. Web data that count the number of clicks
on given sites, or model the usage of different pages are also modeled as
time series.
In the last few years, the advances in mobile computing, sensor and GPS
technology have made it possible to collect large amounts of spatiotempo-
ral data and there is increasing interest to perform data analysis tasks over
this data [4]. For example, in mobile computing, users equipped with mobile
devices move in space and register their location at different time instants
to spatiotemporal databases via wireless links. In environmental informa-
tion systems, tracking animals and weather conditions is very common and
large datasets can be created by storing locations of observed objects over
time. Other examples of the data that we are interested in include features
extracted from video clips, sign language recognition, and so on.
Data analysis in such data includes determining and finding objects that
moved in a similar way or followed a certain motion pattern. Therefore, the
objective is to cluster different objects into similar groups, or to classify an
object based on a set of known examples. The problem is hard, because the
similarity model should allow for imprecise matches.
In general the time-series will be obtained during a tracking procedure,
with the aid of devices such as GPS transmitters, data gloves etc. Here also
athens 1
athens 2
outliers
outliers
time
Fig. 1. Examples of video-tracked data representing 2 instances of the word '
athens
'.
Start & ending contain many outliers.
