AmbiguityVis: Visualization of Ambiguity in Graph Layouts

Area of Science:

• Information Visualization
• Computer Graphics
• Network Analysis

Background:

• Node-link diagrams are intuitive for network exploration but automated layouts often create visual ambiguities.
• Existing graph drawing strategies struggle to satisfy all aesthetic criteria simultaneously, leading to misinterpretations.
• Visual ambiguities in network drawings can hinder the accurate understanding of network structures and relationships.

Purpose of the Study:

• To present a technique for highlighting common visual ambiguities in node-link diagrams.
• To quantify areas of ambiguity using novel and existing metrics.
• To provide visual feedback to developers for improving graph layout algorithms.

Main Methods:

• Developed a technique to detect visual ambiguities: spatial relationships, community overlaps, and edge bundling issues.
• Proposed new metrics for abnormal edge lengths, community overlap, and node/edge aggregation.
• Utilized a heatmap-based visualization to display quantified ambiguities for developer feedback.

Main Results:

• The technique effectively highlights problematic areas in graph layouts that can impede network understanding.
• Novel metrics provide quantitative measures for specific types of visual ambiguities.
• Heatmap visualization offers intuitive feedback for identifying and addressing layout issues.

Conclusions:

• The proposed metrics and heatmap visualization enable users to explore graph layout ambiguities from multiple perspectives.
• This approach facilitates informed decisions in selecting and refining graph layout strategies.
• The technique aids in creating more accurate and interpretable network visualizations.