Toward a Grammar-Based Foundation of Visual Analytics

Abstract

Visual analytics has produced numerous innovative solutions for helping humans gain insight from complex data, yet fundamental questions remain unanswered, including how interactive visualizations support insight generation. We propose grammars as a framework for reasoning about complex, often under-specified concepts at the core of visual analytics. We introduce a context-free grammar inspired by scientific hypotheses that captures the interplay among humans, data, and systems. We demonstrate how grammars define infinite analytical spaces through finite production rules, describe iterative updates and refinements that mimic human analysis, and encode ambiguity through partial specification. By formalizing relationships among data, visualization, and human analysis from a grammar-based perspective, we provide a unified representation for reframing and investigating fundamental questions in visual analytics.