GoFish: A Grammar of More Graphics!

Area of Science:

• Computer Science
• Human-Computer Interaction
• Information Visualization

Background:

• The Grammar of Graphics (GoG) is a foundational theory for visualization grammars like ggplot2 and Vega-Lite.
• While GoG expanded visualization capabilities beyond fixed chart types, it has limitations, excluding charts like mosaics and waffles.
• Current GoG implementations require custom mark types or lower-level frameworks for unsupported charts.

Purpose of the Study:

• Introduce GoFish, a novel declarative visualization grammar.
• Formalize Gestalt principles within a visualization grammar framework.
• Enable recursive composition for a more expressive and flexible design space.

Main Methods:

• Developed GoFish, a declarative grammar formalizing Gestalt principles (e.g., uniform spacing, containment, connection).
• Implemented graphical operators that support arbitrary nesting and overlapping (recursive composition).
• Demonstrated GoFish's capabilities through a diverse gallery of visualization examples.

Main Results:

• GoFish operators enable greater expressive power compared to existing GoG implementations.
• Users can arrange shapes in diverse ways while retaining high-level grammar benefits (scale resolution, coordinate transforms).
• Recursive composition creates an infinite design space, blurring lines between low-level and high-level grammars.

Conclusions:

• GoFish overcomes limitations of current visualization grammars by incorporating Gestalt principles and recursive composition.
• This approach offers an updated theory of visualization, embracing an innumerable space of graphic representations.
• GoFish provides a powerful, flexible, and concise tool for visualization design.