Visualization generation tools automate the visualization design process by generating visualization designs based on visualizaition heuristics such as effectiveness rankings and/or statistical properties within data.
While these tools simplify the design generation process, the designs they generate are not easily modified and are limited to a small subset of visualization designs types. Additionally, generating designs based on
effectiveness rankings and perceptual studies alone may not fully capture the design space of appropriate visualizations for a users design and analytic intent.
My current research examines how we can combine the vast knowledge embedded in existing visualization examples, insights into the behavior patterns of designers, and the latest advancements in AI.
The goal is to develop systems that not only encourage the exploration of diverse design candidates but also facilitate the identification and co-design of expressive visualization designs.
VisAnatomy: An SVG Chart Corpus with Fine-Grained Semantic Labels