slide1 n.
Skip this Video
Loading SlideShow in 5 Seconds..
Node-Attribute Graph Layout for Small-World Networks PowerPoint Presentation
Download Presentation
Node-Attribute Graph Layout for Small-World Networks

Node-Attribute Graph Layout for Small-World Networks

97 Vues Download Presentation
Télécharger la présentation

Node-Attribute Graph Layout for Small-World Networks

- - - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript

  1. Node-Attribute Graph Layout for Small-World Networks Helen Gibson Principal Supervisor: Dr. Paul Vickers 1st Supervisor: Dr. Maia Angelova 2nd Supervisor: Dr.FouadKhelifi Previous Supervisor: Dr. Joe Faith

  2. What is a Graph? Relationships between concepts Mathematics and Graph Theory Information Visualisation Graph Drawing Network Visualisation Network Graph

  3. Examples Social Networks World Wide Web Biological Networks IP Addresses

  4. What’s the Problem? It looks nice but is it doing anything useful? Typical complaint: Giant-Hairball Caused by force-directed algorithms Old, but still popular and most commonly used Connected nodes attract, other repel Yeast interaction network in Gephi

  5. How Can This Be Solved? Node Attributes Example – Social Network Node = People Links = Friendships Attributes = age, gender, location, games they interact with, pages they had liked etc. Typical Usage – As retinal variables Use to tell us more information about the graph Uses beyond retinal variables?

  6. Research Aims Novel graph layout based on node-attributes Many node attributes -> use a dimension reduction technique Visual analysis of graphs Visual Analytics - the science of analytical reasoning facilitated by interactive visual interfaces. [Thomas and Cook, 2005] To further understand the connectivity and structure of the graph

  7. Node-Attributes to Dimensions Attributes as a second set of links Nodes Attributes Each attribute node is a dimension and existence of a link is a value for that dimension on that node

  8. Dimension Reduction and TPP In visualisation: • Many variables form a high-dimensional space reduce to 2 or 3 dimensions that can be seen on a display. • Linear projections Projection Pursuit: • Finds the most ‘interesting’ projection • Interestingness depends on the data Targeted Projection Pursuit (TPP): • Interactive • Searches for a projection closest to a users desired view • In following case, separation of the clusters as far as possible.

  9. Small-World Networks Networks that are: • Highly clustered • Smaller than average shortest path length • An Example: • 4 clusters • Small nodes are attributes Clustering – users’ most valued layout feature

  10. Comparison Force-Directed Graph+TPP

  11. What’s Next? ‘How much better is the clustering?’ Real world domain applications What do we learn about the data from the layout? Evaluation

  12. Publications Gibson, H. (2010) Data-driven layout for the visual analysis of networks. GROUP28: The XXVIII International Colloquium on Group-Theoretical Methods in Physics. Newcastle-upon-Tyne, July 2010. Poster presentation. Gibson, H. , Faith, J. (2011) Node-attribute graph layout for small-world networks. 15th International Conference on Information Visualisation. London, July 2011