Article Preview
TopContext And Motivation
Most OLAP applications are focused on textual data and numerical measures even though available studies have concluded that 80% of data is associated with spatial information (Bédard, Rivest, & Proulx, 2006). Consequently, the integration of spatial data within the multidimensional model was envisaged. Rivest, Bédard, and March (2001) defined the Spatial OLAP (SOLAP) concept as “a visual platform built especially to support rapid and easy spatio-temporal analysis and exploration of data following a multidimensional approach comprised of aggregation levels available in cartographic displays as well as in tabular and diagram displays''. SOLAP systems allow the integration of spatial data, included either in dimensions (spatial dimensions) or in fact tables (spatial measures), in OLAP applications enabling cartographic displays on those applications (Rivest, Bédard, & March, 2001). This way, thematic maps are produced by using the members of spatial dimensions and the numerical measures, combining them with the visual variables (Rivest, Bédard, Proulx, Nadeau, Hubert, & Pastor, 2005).
In this new environment for the analysis of spatial data, several benefits have been mentioned from thematic maps visualization enabled by SOLAP systems. Among them we found better and faster global perception of query results, and the possibility to discover correlations between phenomena, as detailed in Bédard, Rivest, and Proulx (2006).
For the users, it is worth to mention that there is a difference in the results that can be analyzed by an OLAP user compared to a SOLAP user. A typical result of the former involves one to two dozen lines with the aggregated data. However, the latter may involve hundreds of lines if the user is interested in the analysis of data at a lower level of granularity (e.g., customer level). If this is the case, and depending on the geographical distribution and the spatial objects’ representation, a thematic map can easily become cluttered and hard to analyze, as illustrated in the examples displayed in Figure 1.
Figure 1. Examples of cluttered maps
Regarding the example shown in Figure 1a, the points are airports locations and the brightness associated to them is given by the value of a numerical measure. For this map, the user is unable to compare the numerical measure value of different airports as many markers are hidden due to an overlapping between them. The same happens in Figure 1c, where we have pie charts with information of several facilities in a geographical location. In the Figure 1b, the blue markers correspond to the departure airports; the red markers are the arrival airports; and, the value of the numerical measure is given by the arc’s width. Again, the map comprehension/visualization becomes harder due to the high number of arcs and their superposition. Moreover, this last example is a result based on a novel approach to support queries which has involved two spatial attributes from different spatial dimensions (in the query select clause). This novel approach is proposed in this work and further detailed in subsequent sections of this paper.
The presented examples result from the interaction with SOLAP+ system (Jorge, 2009; Silva, 2010), which is afterwards described in this paper. In all of them, the map showed to be inadequate to the user visualization and corresponding analysis.