Pathfinder: Visual Analysis of Paths in Graphs

Area of Science:

• Graph theory
• Information visualization
• Data analysis

Background:

• Path analysis is crucial in many domains but faces scalability challenges with large, multivariate networks.
• Traditional node-link layouts struggle to visualize complex attributes of real-world networks effectively.
• Visualizing rich node and edge attributes exacerbates scalability issues in graph analysis.

Purpose of the Study:

• To present visual analysis solutions for path-related tasks in large and highly multivariate graphs.
• To overcome the scalability limitations of traditional graph visualization methods for path exploration.
• To equip analysts with effective tools for exploring large, attribute-rich networks by focusing on paths.

Main Methods:

• Introduced Pathfinder, a visual analysis technique for querying paths with various constraints.
• Visualized query results as a ranked list displaying rich attribute data and a node-link diagram for topological context.
• Employed strategies like incremental query results to ensure scalability for graphs with tens of thousands of nodes and edges.

Main Results:

• Pathfinder effectively visualizes paths in large, multivariate graphs, addressing scalability issues.
• The system allows ranking paths based on topological properties and attribute-derived scores.
• Demonstrated utility in analyzing coauthor networks and biological pathways.

Conclusions:

• Focusing on paths enables scalable visualization of multivariate graphs.
• Pathfinder provides a powerful tool for analysts to explore and understand complex network structures and attributes.
• The technique enhances the analysis of large-scale networks across various scientific domains.