Preserve high-importance regions when creating visualization thumbnails

Build the thumbnail from the most important regions instead of shrinking the full visualization uniformly. For example, preserve the title, main supporting text, and extreme data regions, and remove lower-importance rows or columns to reach the target aspect ratio.

Tiny full-image thumbnails often make text and key data regions too small to recognize quickly. A thumbnail that keeps the regions people inspect first gives readers a faster summary when they scan a grid of results.

Mechanism: Preserving high-importance text and key data regions makes the thumbnail easier to identify and match to a search target.

Evidence: In the paper’s search task, participants found the described visualization in fewer clicks with importance-based thumbnails than with uniformly resized originals, and the retained thumbnail content typically included titles, supporting text, and data extremes (Bylinskii et al., 2017).

• User Goal: Help readers find the right visualization in a large collection.
• Task: Summarize each visualization into a small preview for scanning.
• Data: Dense data visualizations with important text and graphical regions.
• Chart Setting: Single-view visualizations displayed as small thumbnails in a grid or list.
• Audience: Readers scanning many candidate results.
• Success Criterion: Readers identify the target visualization in fewer clicks.

Break it when: The importance map is highly non-uniform within text or graphical elements. Why: The paper notes that this can cut off parts of elements or text in downstream thumbnailing.

Sacrifice: Full context and exact original layout. Risk: Important text or marks can be cut at the boundary of the kept region. Mitigation: Soften retained boundaries with a fade to white after carving.

Mistake: Uniformly resize the full visualization into a tiny thumbnail. Why it fails: Key text and high-value regions become too small to support fast retrieval.

Failure Sign: Readers need several clicks to find the target visualization in a thumbnail grid. Quick Check: Compare the importance-based thumbnail side by side with a simple resized version and inspect whether the title and key supporting text remain visible. Stronger Test: Run an A/B search task and compare average clicks to the target between the two thumbnail versions.

• Remove low-importance rows or columns until the thumbnail reaches the target proportions.
• Preserve the title and main supporting text inside the retained region.
• Keep the data extremes that remain visually diagnostic at small size.
• Fade carved boundaries to white so the remaining content reads as a coherent thumbnail.