Geoscience Reference
In-Depth Information
inherent among the geographic phenomena being represented. The ability
to abstract data being ever more important in today's information society -
in which the volume of data exceeds our insatiable appetite for more
(Mackaness et al. 2007).
Already Eckert (1921) began to consider the necessity of cartographic
generalization. Further theoretical background of cartographic generaliza-
tion of map objects can be found in Bjørke (1996), Mustière and Moulin
(2002) and Hake et al. (2002). In this approach for point selection the
Polarization Transformation (PT) is used and for each point a global and a
local characteristic is defined.
Research in this field draws on expertise in exploratory data analysis or
visual analytics, interface design, agent-based methodologies and cognitive
ergonomics. Some of our own pre-studies within interactive visual tools
for analyzing point datasets can be found in (Peters and Krisp 2010, Krisp
and Peters 2010, Krisp et al. 2010 and in Krisp et al. 2009). Investigations
in this paper may support visual analytics approaches currently examined
in a number of research projects, e.g. VISMASTER (2010) and NVAC
(2010).
Bertin (1983) believed that visualization is not effective unless it allows an
immediate extraction of the essential information. Many traditional data
visualization techniques which proved to be supportive for exploratory
analysis of datasets of moderate sizes fail when applied to large datasets.
According to Andrienko and Andrienko (2007), two approaches can han-
dle with huge data sets. One approach is data aggregation which considers
the clusters instead of the original data. The second approach is data selec-
tion which focuses on a portion of characteristic data items. Unfortunately,
none of the two approaches can satisfy the needs of exploratory data
analysis. These needs are described in Bertin (1983): Exploratory data
analysis requires a consideration of the data on all levels: overall (consid-
ering a dataset as a whole), intermediate (viewing and comparing collec-
tive characteristics of arbitrary data subsets, or classes), and elementary
(accessing individual data items). Andrienko and Andrienko (2007) sug-
gested therefore a combination of data aggregation and data selection, i.e.
to show the entire data set and arbitrarily defined subsets in an aggregated
way. They provide a solution using an adapted parallel coordinate plots. In
this work an interactive tool is developed, which enable the user to per-
form point data selection.
