Design-Specific Transformations in Visualization

TL;DR

The paper addresses how to reason about visualizations by distinguishing design-specific transformations from visual encoding and by modeling the user task as a function over data. It introduces a Transform-centric Model that extends the InfoVis Reference Model by representing visual mapping as $V = e(f(D))$ with $f()$ as design-specific transformations and $e()$ as encoding, thus treating tasks as compositions over $D$. Key contributions include the No Free Lunch conjecture, a data- and view-proxy framework for evaluating task performance, and a cost-model approach to visualization evaluation and experiment design. These ideas enable clearer reasoning about visualization correctness and effectiveness, inform experiment design, and impact visualization theory by linking data transformations, encodings, and user tasks.

Abstract

In visualization, the process of transforming raw data into visually comprehensible representations is pivotal. While existing models like the Information Visualization Reference Model describe the data-to-visual mapping process, they often overlook a crucial intermediary step: design-specific transformations. This process, occurring after data transformation but before visual-data mapping, further derives data, such as groupings, layout, and statistics, that are essential to properly render the visualization. In this paper, we advocate for a deeper exploration of design-specific transformations, highlighting their importance in understanding visualization properties, particularly in relation to user tasks. We incorporate design-specific transformations into the Information Visualization Reference Model and propose a new formalism that encompasses the user task as a function over data. The resulting formalism offers three key benefits over existing visualization models: (1) describing task as compositions of functions, (2) enabling analysis of data transformations for visual-data mapping, and (3) empowering reasoning about visualization correctness and effectiveness. We further discuss the potential implications of this model on visualization theory and visualization experiment design.

Figures (7)

• Figure 1: The proposed model in this paper differs from the Infovis Reference Model in two important ways: we decompose visual mappings into design-specific transformations (e.g., stacking, quantization, calculating statistics) from visual encoding, and we model the task $q(D)$ as a function over the input dataset that the user wishes to estimate.
• Figure 2: Summary of the Transformation-Centric Model. An analytic task is defined as a query over $D$, and can be answered by a query over the prepared table $P$ or by the user using a rendering of the visual abstraction $V$.
• Figure 3: Point marks that render counts by $g$.
• Figure 4: New York Times visualization that tells a data story about the post-pandemic surge in child migrants.
• Figure 5: WireVis system to detect fraudulent banking activity.

Theorems & Definitions (10)

• Example 1
• Example 2
• Example 3
• Example 4
• Example 5
• Conjecture 1: No Free Lunch
• Example 6: Pie charts and "parts-to-whole" relations
• Conjecture 2: Visualization Utility is Task-dependent
• Example 7
• Example 8