Understanding Data-Driven Visual Encodings through Deconstruction

Our ongoing work aims to better understand the relationship between visual marks and data encodings that comprise innovative data visualizations. By innovative, we refer to data visualizations that in part extend or deviate from conventional visualization techniques. Authors of these types of visualizations engage and inform their audience by incorporating novel data-bound compositions. In our work, we deconstruct a set of these visualizations into their elemental data-driven marks by employing a semi-structured coding process. This process serves as the initial step for defining a visual grammar that can act as the building blocks for creative data visualizations. The primary purpose of this model is to define the set of graphics, data-bindings, and visual styles that researchers may then incorporate into a data visualization authoring tool. Here, we present our analysis approach to deconstruct the set of visualizations and how we cataloged the coded data for dissemination among the visualization community. We then highlight interesting trends observed within the collection of visualizations analyzed thus far.