On Modeling and Analysis of Multidimensional Geographic Databases - Data Warehousing Design and Advanced Engineering Applications - page 105

Figure 4. GeWOlap's user interface

applet developed using MapXtreme Java (MapX-
treme) to support map visualization/interaction
and GIS functionalities. The client tier allows a
visual representation of spatio-multidimensional
structures [(geographic) dimensions and mea-
sures] through pivot table, graphic displays and
interactive map.
Figure 4 displays the visual interface of Ge-
WOlap for the application of figure 2. The pivot
table represents pollution values for Ile de France
region and for its departments (Essone, Hautes de
Seine, etc.). The cartographic component shows a
thematic map representing departments' pollution
values using pie charts.
Spatio-multidimensional and GIS operators are
accessible through the simple interaction with the
pivot table and the interactive map of the client
using only few mouse clicks.
In particular, GeWOlap provides a set of
drill operators which are available through the
interaction with the pivot table and the map:
“roll-up drill-down replace”, “drill-down posi-
tion”, “expand-all”, “drill-through” (Bimonte et
al., 2007a).
For instance, let us suppose that the user wishes
to “see” the measures for departments of the Ile
de France region. By pointing the mouse on that

region she/he can apply the drill-down position
operator (figure 5a). As a result, the pivot table
displays average pollution values for the Ile de
France region, and for other departments (Figure
5b). Several synchronization problems rise from
the topological inclusion relationships between
spatial members of different levels, and the num-
ber of measure values that must be displayed.
For example, unlike the pivot table, the map
cannot display at the same time a region and its
departments, and visualize the pollution values
for each pollutant granting a good cartographic
readability.
The user can cut the hypercube by using the
“Cube Navigator” tool provided by JPivot. The
Cube Navigator provides a tree representation of
dimension members which can be used to custom-
ize pivot table axes and select a sub-set of members.
Moreover, GeWOlap extends OLAP cut opera-
tors by introducing two new cut operators: “Slice
Predicate” and “Slice Position” (Bimonte et al.,
2007b). Thanks to “Slice Predicate”, the user can
select spatial members by directly clicking on the
interactive map. “Slice Predicate” allows cutting
the hypercube through Spatial SQL queries.
Let us suppose the user is interested in pollu-
tion values for departments crossed by from the