A Visual Language for Nested Visualization Design

Abstract

Significant progress has been made in the field of information visualization. Many programming libraries (like D3) enable the creation of almost any 2D visualization. However, this power of expression is not available for non-coders. In this thesis we show how non-coders can create both simple and complex visualizations using only drag & drop operations. We define a data structure (Visception Tree) that can represent arbitrarily nested and layered charts. With this data structure we define a hierarchy or visualizations embedded within one another, or layered side by side, or a combination of the two. Each chart can be edited separately and intuitively, and selecting the chart is done in an outline view - similar to a file view. Such simple ways of interaction are made possible by the Visception Tree being easily mappable to user interface actions as well as being flexible enough to encapsulate arbitrary hierarchies and layerings of charts. The viability of these ideas is demonstrated by showing how some complex visualizations can be made with just a few drag & drop operations - enabling the creation of visualizations in just a few minutes.