Download presentation
Presentation is loading. Please wait.
1
Network Visualization & Interaction
Jay Thom
2
Outline Introduction Background Related Works Methodology Conclusion
Why is this important? Background History Related Works Problems Solutions Methodology What I will contribute Conclusion
3
Why Is This Important? My motivation: Internet Topology Measurement
Our project Dealing with large amounts of data How do we interpret it? Introduction * Background * Related Works * Methodology * Conclusion
4
Generating Massive Data in Daily Life…
Finance Healthcare Sensor Networks Power Grid National Security Consumer Data 50,000 GB per second! Data is meaningless unless we can interpret it! Introduction * Background * Related Works * Methodology * Conclusion
5
How Humans Assimilate Data…
Introduction * Background * Related Works * Methodology * Conclusion
6
Much Data Representable as Networks
We need to interpret data to make it useful: Automated algorithms Data mining Machine learning Human domain expertise needed for optimization: Data needs to be interpreted quickly Inconsistencies need to be identified We need to combine automation with interaction Introduction * Background * Related Works * Methodology * Conclusion
7
History of Data Visualization
Charles Joseph Minard ( ), Napoleon’s Grand Army in Russia Introduction * Background * Related Works * Methodology * Conclusion
8
Modern Network Graph Visualizations
The Internet Introduction * Background * Related Works * Methodology * Conclusion
9
Problems…Readability
Ben Shneiderman and Cody Dunne; ”Interactive Network Exploration to Derive Insights” Readability Metrics: Crossing number number of times an edge crosses another edge Drawing area (smallest bounding box) smallest possible bounding box looks best Symmetry groups find symmetry in the graph and work around it Minimal bends in edges keep edges as straight as possible Minimal edge lengths Minimize total length of edges Angular resolution higher angle resolution is desirable Slope number use minimal number of distinct edge slopes Introduction * Background * Related Works * Methodology * Conclusion
10
Readability - Motifs N. Karracher et al: ”Visual Techniques to Support Exploratory Analysis of Graph Data” Cliques Connectors Fans Introduction * Background * Related Works * Methodology * Conclusion
11
Readability - Motifs Introduction * Background * Related Works * Methodology * Conclusion
12
Problems…Layout Yifan Hu; “Algorithms for Visualizing Large Networks”
Not all graph can be aesthetically embedded into two or three-dimensional space small world graphs Graphs exhibiting power law degree distribution * Graphs of this nature require an interactive system to comprehend and explore Spring-Electrical Model Force-directed Quad-tree Stress-Strain Models Stress model Strain model (classical MDS) Multi-scale algorithm Landmark MDS Pivot MDS Introduction * Background * Related Works * Methodology * Conclusion
13
Layout – Spring Electrical Models
Force-Directed: Highly scalable, can lay out millions of nodes in minutes Does not cope well with pre-defined edge lengths Introduction * Background * Related Works * Methodology * Conclusion
14
Layout – Spring Electrical Models
Quad-Tree: Simulates Force-Directed model Useful when there are vertices far from the main body of the group Introduction * Background * Related Works * Methodology * Conclusion
15
Layout – Stress/Strain Models
Stress Model: Better than Force-Directed for pre-defined edge-lengths Spring forces attach each node to its neighbors Ideal spring length is pre-defined Introduction * Background * Related Works * Methodology * Conclusion
16
Layout – Stress/Strain Models
Pre-dates the Stress model Fits to the inner product of positions Centers nodes around an origin Introduction * Background * Related Works * Methodology * Conclusion
17
Problems…Interaction
Question: What is interaction? Ji Soo Yi et al; “Toward a Deeper Understanding of the Role of Interaction in Information Visualization Interaction Categories: Select Explore Reconfigure Encode Abstract/Elaborate Filter Connect Introduction * Background * Related Works * Methodology * Conclusion
18
Interaction - Select “Mark something as interesting”
Allows user to mark a data item to keep track of it Item is highlighted or labeled, and can be easily located after rearrangement Preceding action to subsequent operations Introduction * Background * Related Works * Methodology * Conclusion
19
Interaction - Explore “Show me something else”
Rearrange the view to bring out different details Move smoothly from one view to another Pan or zoom a view to enhance detail Introduction * Background * Related Works * Methodology * Conclusion
20
Interaction - Reconfigure
“Show me a different arrangement” Change the spatial arrangement of representations Reveal hidden characteristics of data Provide different perspectives Introduction * Background * Related Works * Methodology * Conclusion
21
Interaction - Encode “Show me a different representation”
Change the way data is represented Uncover new aspects of relationships between data points Change colors, shapes, sizes, chart/graph types Introduction * Background * Related Works * Methodology * Conclusion
22
Interaction – Abstract/Elaborate
“Show me more or less detail” Adjust the level of abstraction Alter representation from overview to detailed view of individual cases Show sub-trees of a tree Show tool-tip view (example) Introduction * Background * Related Works * Methodology * Conclusion
23
Interaction - Filter “Show me something conditionally”
Enable user to change set of data items being presented based on some condition Show only data of a certain range Change color of filtered data so that it is still visible but discernable from non-filtered data Introduction * Background * Related Works * Methodology * Conclusion
24
Interaction - Connect “Show me related items”
Highlight associations and relationships between items Show hidden items relevant to a specified item Hover over an item to reveal hidden relationships Click on item to show item details not visible Introduction * Background * Related Works * Methodology * Conclusion
25
U.S. Senators by Voting Similarity 2014
Introduction * Background * Related Works * Methodology * Conclusion
26
U.S. Senators by Voting Similarity 2014
Introduction * Background * Related Works * Methodology * Conclusion
27
Visualization Tools Open-Source Graph Visualization Tools: Guess Gephi
NodeXL Cytoscape NetworkX iGraph SNAP Sigmajs D3.js Linkurious R Large Graph Layout (LGL) Cuttlefish Graphistry Graph-tool libSNA MeerKat Netlytic NetworKit Pajek SocNetV Socioviz Statnet SUBDUE Tulip Visone GraphChi Graphviz JUNG Introduction * Background * Related Works * Methodology * Conclusion
28
Categorization User interface Programming language
Ease or difficulty to use Range of tools available Range of interaction available Visual quality Limitations on number of nodes/edges Overall performance Introduction * Background * Related Works * Methodology * Conclusion
29
Conclusion Why is this important? History Problems Solutions
What I will contribute Introduction * Background * Related Works * Methodology * Conclusion
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.