Pattern-Driven Navigation in 2D Multiscale Visualizations with Scalable Insets.

Area of Science:

• Information Visualization
• Human-Computer Interaction
• Data Science

Background:

• Exploring numerous annotated patterns in multiscale visualizations is difficult due to changing visual representations and lost context during zooming.
• Navigation in large datasets is often time-consuming and inefficient.

Purpose of the Study:

• To introduce Scalable Insets, a novel technique for interactively exploring and navigating large numbers of annotated patterns in multiscale visualizations.
• To address challenges in pattern identification, comparison, and contextualization across different zoom levels.

Main Methods:

• The Scalable Insets technique visualizes small annotated patterns using magnified thumbnail views (insets).
• Insets are dynamically placed within or along the viewport boundary to balance locality and context.
• Annotated patterns are interactively clustered by location and type, represented as aggregated insets for scalable exploration.

Main Results:

• A user study showed Scalable Insets speeds up visual search and improves pattern comparison accuracy.
• The technique may lead to slower frequency estimation compared to baseline methods.
• Expert feedback indicated Scalable Insets is easy to learn and useful for open-ended data exploration.

Conclusions:

• Scalable Insets offers an effective approach for navigating and analyzing large annotated datasets in multiscale visualizations.
• The technique shows promise for applications in fields like genomics, facilitating quicker data insights.