VisPuzzle: Task-Aware Composite Visualization Construction
Authors
Zheng Wang (Tsinghua University), Zhiyang Shen (Tsinghua University), Lingyun Yu (Xi'an Jiaotong-Liverpool University), Shixia Liu (Tsinghua University)
Presentation
- Session
- I feel tangled in a net
- Time
- Wednesday, Nov 11, 08:24 – 08:36 (US/Eastern) · session 08:00 – 09:30
- Location
- Hall America center
Links
Sign in to access the preprint PDF.
Sign in- Download Supplemental Material
Keywords
Composite visualization, Monte Carlo Graph Search, data insight
Abstract
Compositing multiple visualizations into a coherent whole remains challenging due to the vast design space and the need to balance the coverage of task-relevant data insights (e.g., trends and outliers), perceptual clarity, and aesthetic quality. In this paper, we present VisPuzzle, a task-aware method that formulates visualization composition as a stepwise search problem over a composition graph. In this graph, nodes represent either data composition operations (e.g., union, join) or visual composition operations that determine component relationships, spatial arrangements, or component proportions, and edges encode feasible transitions between operations. We employ Monte Carlo Graph Search to efficiently identify high-quality composition candidates from this graph, guided by a reward function that balances task relevance, perceptual effectiveness, and aesthetic coherence. A use case and a user study show that the top-ranked candidates produced by VisPuzzle align closely with human judgments of composition quality, demonstrating its utility in supporting principled and scalable visualization composition.
For Practitioners
Data analysts, data journalists, and visualization designers who need to communicate multiple related data insights would be interested in this work. They can apply VisPuzzle to efficiently generate and compare task-relevant composite visualization designs that balance task relevance, perceptual effectiveness, and aesthetic coherence.