graph visualization using hierarchical aggregation and edge bundling
TRANSCRIPT
Se#ng
• Social bookmarking dataset • URLs described by tags • Dataset characteris9cs: • 61,665 posts • ~430,000 triples • Nov 1 – Nov 30, 2009
Goal
• Facilitate exploratory search for URLs by topics of interest
• Show distribu9on of dataset at different levels of granularity
• Enable to inves9gate groups of interests – What groups of interests exist? – Are they somehow related? – How do they evolve over 9me?
Our Approach
Graph Visualiza-on using Hierarchical Aggrega-on and Edge Bundling
Why?
• Edge bundling methods proved useful for reducing cluVer
• Hierarchical techniques provide basis for naviga9on and level of detail rendering
Visual Representa9on
• Planar graphical representa9on – variable magnifica9on – Hierarchical structure to zoom back and forth – Edge Bundeling on each hierarchy level to reduce cluVer
• Nodes can be selected via mouse click and edges connec9ng to other clusters are highlighted
Take Away Messages
• Scalable graph viz • Reveals groups of interest expressed by user generated tags
• Hierarchical graph viz layout enables analysis at different levels of detail
• Edge bundeling reduces cluVer • Next step: Inves9gate how clusters evolve over 9me
Vedran Sabol
vsabol@know-‐center.at
Elisabeth Lex
Our Team
Ralph Wozelka
rwozelka@know-‐center.at