A software architecture for automated information visualization

Information visualization is the application of computer-aided visualization technology to data that is not inherently visual in nature. Information visualization systems usually support very specific tasks (e.g., visualization of code execution, parallel code simulation, relational database visualization). Whereas there exist general-purpose systems for scientific visualization, there are no general-purpose "information visualization" systems. To address this requirement, the Intelligent Visualization System (IVS) is presented for general information visualization. Its underlying architecture enables the automatic generation of visualizations without specific directions regarding how to depict the visualization. This approach frees both the programmer and the user from the task of deciding the most effective way of showing information visually. This dissertation presents the architecture of the IVS and discusses the results from a prototype implementation. In the process of developing the architecture, a unique taxonomy of the visual presentation of information was developed and utilized. Projected utility of a visualization is modeled using a set of visual metrics. User evaluations of a visualization can be used to tune the application of these metrics. Implementation of the IVS prototype highlighted several ways in which the field of visualization must mature to allow an IVS system to easily integrate into existing applications.