Towards a generative theory of diagram design

We describe the theoretical background for AVE, an automatic visualization engine for semantic networks. We have a functional notion of aesthetics and therefore understand meaningfulness as a central issue for information visualization. This implies that the diagrams should communicate the characteristics of the data as effectively as possible. In this generative theory of diagram design, we include data characterization, systematic use of graphical means of expression and the combination of graphical means of expression. After giving a brief introduction and an application scenario we discuss these aspects in detail. Finally, a process model of an automatic visualization process is sketched and directions for further research are outlined.