Experimentally driven visual language design: texture perception experiments for iconographic displays

Abstract

Visualization researchers of the Exploratory Visualization (Exvis) project are studying the representation of multidimensional databases as two-dimensional arrays of data-driven icons. Each data point in n-dimensional space is converted into one icon in the display; both visual and auditory features of the icon are determined by the data. The display's texture is produced by packing large numbers of small icons together so densely that they lose their individual identities. The premise of the technology is that interesting features in the visual and auditory texture of an iconographic display will point to interesting features in the data. The technology is in the early stages of formal study. The short-term goal is to provide a workstation that will enable a researcher who is neither a programmer nor a trained experimentalist to design, implement, conduct and analyze human factors experiments for studying the iconographic data-display technique. The long-term goal is to use the information gathered from such experiments to provide a powerful data-representation language for scientists to use for visualizing large data sets.<>