Special Interest Group on Visualization Grammars

Abstract

The popularity of these visualization grammars is evident. As open-source packages, ggplot2 (R) has around 1.5 million down-Visualization grammars, often based on the Grammar of Graphics, are popular choices for specifying expressive visualizations and loads from CRAN per month 1 and Vega-Lite (JSON specifcation) supporting visualization systems. However, there are still open has 2 million CDN hits per month 2 . Users of visualization grammars are well beyond the visualization research community -journalquestions about grammar design and evaluation not well-answered in visualization research. In this SIG, we propose to discuss what ists use D3 3 and ggplot2 [5] in their reporting, and authors of ggplot2 books have background in felds like statistics [19] and makes a grammar "good" and explore evaluation methodologies sociology [4]. best suited for visualization grammars. Increasingly, visualization grammars have become building blocks for other formalisms and systems. To support data explo-CCS CONCEPTS ration and question answering in visual analysis, Voyager 2, a • Human-centered computing → Visualization design and mixed-initiative system, is powered through CompassQL, "a generevaluation methods; Visualization theory, concepts and alization of the Vega-Lite grammar" [14, 22]. To depict probability paradigms; Visualization application domains. distributions, A Probabilistic Grammar of Graphics makes probability expressions such as P(A|B) frst-class citizens on top of the KEYWORDS original GoG [13]. For high-performance visual analytics, P6 lever-Grammar of Graphics ages a declarative grammar to integrate interactive visualizations

ComputableViz: Mathematical Operators as a Formalism for Visualization Processing and Analysis

“…We decide our scope of operation targets by selecting visualization primitives in declarative visualization grammars. Specially, we refer to the Vega-Lite [71] language which has been gaining extensive popularity [64]. At the top level, a Vega-Lite specification consists of view specification, data/dataset, transform, mark, encoding, view composition, parameters, and configurations.…”

No Grammar to Rule Them All: A Survey of JSON-style DSLs for Visualization

“…Despite its prevalent use, the term "visualization grammar" is not well defined and is used in a variety of ways [89]. This term is sometimes used in the generative syntactic sense [82], referring to a system of rules that can be repeatedly applied to create particular shades of meaning.…”

“…We expand upon these studies through a more in-depth survey of narrower scope. Pu et al [89] highlight the pressing need for more formal study of these entities. We seek to explore and address the questions they raise, as well as support future work by developing a richer understanding of the state of the art of this language form.…”

“…Domain-specific languages (DSLs) represented in standard data serialization formats, such as JSON or YAML, are an increasingly common [89] interface for the specification of visualizations across an array of contexts and tasks. These restricted textual languages allow for the declarative specification of both static and interactive graphics in a systematic manner that can be manipulated both by humans, making them attractive for end-user programming, and computational agents, making them appealing for artificial intelligence applications [123].…”

“…2) of this approach, there has been little detailed analysis of the characteristics of these systems. Pu et al [89] highlight the need for additional study of visualization grammars, while Wongsuphasawat [117] surveyed the more general space of JavaScript (JS) visualization libraries. Although they are insightful, these works leave critical questions about these DSLs open: What problems do they seek to solve?…”