On Geometry and Transformation in Map-Like Information Visualization

This paper proposes to discuss particular projection techniques, like multidimensional scaling or self-organizing maps, with reference to the geometric primitives they employ, which will drive the choice of geometric and symbolic transformations that are necessary to achieve a particular visualization.

Abstract

A number of visualization techniques have been put forward that implement a map metaphor to display abstract, non-georeferenced information. This paper refers to these as map-like information visualizations that are distinguished from other information visualization approaches in a number of ways. It interprets some of the principles underlying these techniques within a framework informed by geographic information science (GIScience). Recent geographic efforts in this research area have linked ideas about the nature of geographic information to cognitive schemata proposed by cognitive linguists. This paper draws on the arguments that have emerged from those efforts regarding the nature and usefulness of geographic metaphors. It proposes to discuss particular projection techniques, like multidimensional scaling or selforganizing maps, with reference to the geometric primitives they employ. These primitives will drive the choice of geometric and symbolic transformations that are necessary to achieve a particular visualization. Designers of map-like visualizations are thus challenged to seriously consider the implications of particular computational techniques and the consequences of symbolization choices.