Databases Reference
In-Depth Information
Figure 3.14:
The wine dataset scatterplot matrix.
It is the authors' belief that most of the available Infovis systems fail to address, in some
way, these basic issues, resulting in systems that do not allow exploitation of all the underlying
representation alternatives.
Space limitation refers to the problem of presenting the abstract representation on a real
screen, taking into account screen size, collisions, user perception, etc. Several techniques have been
developed to deal with such issues, and we discuss here the guidelines underlying them, providing
some examples of their application.
It is possible to classify the available strategies into two main classes: the first one foresees
drawing strategies that allow for mapping the representation onto the screen (e.g., providing zoom
and pan mechanism), while the second one encompasses several algorithmic techniques that either
try to ameliorate the image thereby reducing the clutter (e.g., sampling the dataset) or providing
guidance to the end user to easily locate the most interesting visualizations (e.g., automatically
selecting within a large scatterplot matrix the three most representative bi-dimensional scatterplots).
The devised strategies must scale to large datasets: as an example, consider the simple strategy
of mapping a large image on the screen using zooming and scrolling; this can quickly result in
