1 / 45

Visualization Taxonomies and Techniques Graphs

Visualization Taxonomies and Techniques Graphs. University of Texas – Pan American CSCI 6361, Spring 2014. Graphs and Networks. Graphs Show “Connections”.

vidal
Télécharger la présentation

Visualization Taxonomies and Techniques Graphs

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Visualization Taxonomies and TechniquesGraphs University of Texas – Pan American CSCI 6361, Spring 2014

  2. Graphs and Networks

  3. Graphs Show “Connections” NSFNET Traffic 1991, NSFNET backbone nodes are shown at the top, regional networks below, traffic volume is depicted from purple (zero bytes) to white (100 billion bytes), visualization by NCSA using traffic data provided by the Merit • Connections among …anything • Model connected set as a Graph • US telephone system • World Wide Web • Distribution network for on-line retailer • Call graph of a large software system • Semantic map in an AI algorithm • Set of connected friends • Graph/network visualization is one of the oldest and most studied areas of visualization

  4. Graphs Show “Connections” Trade among countries. There are many challenges … • Connections among …anything • Model connected set as a Graph • US telephone system • World Wide Web • Distribution network for on-line retailer • Call graph of a large software system • Semantic map in an AI algorithm • Set of connected friends • Graph/network visualization is one of the oldest and most studied areas of visualization

  5. Graphs Show “Connections” SEMNET, 1987 • Connections among …anything • Model connected set as a Graph • US telephone system • World Wide Web • Distribution network for on-line retailer • Call graph of a large software system • Semantic map in an AI algorithm • Set of connected friends • Graph/network visualization is one of the oldest and most studied areas of visualization

  6. Graphs Show “Connections” vizster, social network, Facebook http://vis.stanford.edu/jheer/projects/vizster/ (download *.wmv) • Connections among …anything • Model connected set as a Graph • US telephone system • World Wide Web • Distribution network for on-line retailer • Call graph of a large software system • Semantic map in an AI algorithm • Set of connected friends • Graph/network visualization is one of the oldest and most studied areas of visualization

  7. Social Network Visualization • Social Network Analysis • Among first studied • http://www.insna.org • Early, by social scientists • Sociologists, anthropologists • Now, very keen interest in social networks • From Facebook to terrorists

  8. Graphs and NetworksSome techniques • Graph layout • Node link layouts • Layered / Sugiyama • Force directed • Other • Matrix layouts • Attribute based layouts

  9. About Graphs • Graph: G = (V, E) • Vertices (nodes) connected by edges (links) • Can have cycles • Edges can be directed or undirected • Degree of vertex is number of edges connected to it • In-degree and out-degree for directed graphs • Edges can have values (weights) • nominal, ordinal or quantitative • Trees • Special case of general graph – no cycle • Typically directed edges • Special designated root vertex

  10. Graph Visualization Challenges • Graph layout and positioning • Make a concrete rendering of abstract graph • Navigation/Interaction • How to support user changing focus, moving around the graph, … • Scale • Small graphs are not hard for above • BUT, 10 – 100 – 1000 … which are the interesting ones • Layout – an entire research community focus

  11. Aesthetic ConsiderationsHow to lay out a graph • Line (edge) Crossings – • minimize towards planar • Total Edge Length – • minimize towards proper scale • Area – • minimize towards efficiency • Maximum Edge Length – • minimize longest edge • Uniform Edge Lengths – • minimize variances • Total Bends – • minimize orthogonal towards straight-line • All at once! • Various studies examined which of the aesthetic factors matter most and/or what kinds of layout/vis techniques look best • Results mixed: Edge crossings do seem important

  12. Graph VisualizationTask Taxonomy • 1. Topology-based tasks • Adjacency: Find the set of nodes adjacent to a node • Accessibility: Find the set of nodes accessible to a node • Common connection: Given nodes, find the set of nodes connected to all • Connectivity: Find shortest path, Identify clusters, Identify connected components • 2. Attribute-based tasks • For nodes: Find the nodes having a specific attribute value • For edges: Given a node, find nodes connected only by certain kinds of edges • 3. Browsing tasks • Follow path: Follow a given path • Revisit : Return to a previously visited node • 4. Overview task • Compound exploratory task : Estimate size of a network, find patterns, …

  13. Layout TechniquesQuick Look • Layout algorithms can create: • polyline edges • planar – • no edge crossings • orthogonal – • horizontal and vertical lines/polylines • grid-based - • vertices, crossings, edge bends have integer coords • curved lines • hierarchies • circular • ... • P. Mutzel, et al. Graph Drawing ’97

  14. Layout TechniquesQuick Look • Will see a couple • Common techniques: • Hierarchical • Force-directed • Circular • Geographic-based • Clustered • Attribute-based • Matrix

  15. Another Graph Drawing ExamplesHuman Disease Lens to view http://www.nytimes.com/interactive/2008/05/05/science/20080506_DISEASE.html

  16. Hierarchical Graph Layout

  17. Hierarchical Graph Layout Sugiyama layout • Often called Sugiyama layout • Try to impose hierarchy on graph • Reverse edges if needed to remove cycles • Introduce dummy nodes • Put nodes into layers, or levels • Order l->r to minimize crossings

  18. Hierarchical Layout Sugiyama Layout Unix “ancestry” • Readable top down flow • Good for graphs that have an intrinsic ordering • Not suitable for graphs that don’t have an intrinsic top down structure • ‘Depth’ in graph mapped to one axis • Lots of gd libs • graphviz lib: http://www.graphviz.org • http://gephi.org

  19. Hierarchical Layout Sugiyama Layout • Readable top down flow • Good for graphs that have an intrinsic ordering • Not suitable for graphs that don’t have an intrinsic top down structure • ‘Depth’ in graph mapped to one axis • Lots of gd libs: • graphviz lib, http://www.graphviz.org • http://gephi.org

  20. Force-Directed Layout

  21. Force-Directed Layout http://mbostock.github.io/protovis/ex/force.html • Define through equations • Spring model (common) • Edges – Springs (gravity attraction) • Vertices – Charged particles (repulsion) • Equations for forces • Iteratively recalculate to update positions of vertices • Seeking local minimum of energy • Sum of forces on each node is zero

  22. Force-Directed Example “Springs (forces) find iteratively find equilibrium”

  23. Force-Directed ExamplesProtovis and D3 Protovis: http://vis.stanford.edu/protovis/ex/force.html D3 (cf collapsible force directed): https://github.com/mbostock/d3/wiki/Gallery

  24. GraphsForce Directed Layout • Very flexible, aesthetic layouts on many types of graphs • Can add custom forces • Relatively easy to implement • Repulsion loop is O(n2) per iteration • Can speed up to O(n log n) using quadtree or k-d tree • Prone to local minima • Can use simulated annealing

  25. GraphsForce directed layout • Many variations, but physical analogy: • Repulsion : fR(d) = CR * m1*m2 / d2 • m1, m2 are node masses • d is distance between nodes • Attraction : fA(d) = CA * (d – L) • L is the rest length of the spring • i.e. Hooke’s Law • Total force on a node x with position x’ • Σ neighbors(x) : fA(||x’-y’||) * (x’-y’) + -fR(||x’-y’||) * (x’-y’) • Examples • 23 second example: http://www.youtube.com/watch?v=AYrkWSDkfLM • 60 second example: http://www.youtube.com/watch?v=QlXRapQW4q0

  26. GraphsForce-directed layout Recall

  27. Force Directed with Magnets http://www.youtube.com/watch?v=K4GOxJywB-U Not much 1st minute

  28. Other Layouts • Orthogonal • Good for UML diagrams • algorithmically complex

  29. Circular Layout • Circular Layout • Very simple • Space vertices out around circle • Draw lines (edges) to connect vertices • But, aesthetic heuristics … • Textarc (more next time)

  30. Textarc: http://www.textarc.org/

  31. Nested Layouts Recursively apply layout algorithms Good for graphs with hierarchical structure

  32. Graphsvisual complexity http://www.visualcomplexity.com/vc/

  33. GraphsAdjacency Matrix

  34. GraphsAdjacency Matrix • Alternative to node link • Adjacency matrix representation • “Mark” where edges are • E.g., A-B, A-C (and inverse)

  35. GraphsAdjacency matrix • Good for dense graphs • Visually scalable • Can spot clusters • Nodes of high degree have many connections, so many entries in adjacency tale • Lots of dots at clusters • However • Abstract visualization • Hard to follow paths

  36. Matrix Representations Interest in matrix representations of graphs Regularity, symmetry, and structure of a matrix good Well understood Difficulties of scale

  37. MatrixExplorer • Provides matrix view in combination with node-link and various operations for gaining different perspectives • Henry & Fekete TVCG (InfoVis) ‘06

  38. Node Reordering Important operation with matrix representations

  39. NodeTrix • Hybrid of matrix and node-link • Henry & Fekete TVCG (InfoVis) ‘07

  40. GraphsAttribute Driven

  41. GraphsAttribute-driven layout Barsky, 2008 • Large node-link diagrams can be challenging to perceptually order • Can use data attributes to perform layout • E.g., scatterplot based on node values • Dynamic queries and/or brushing can be used to enhance perception of connectivity

  42. GraphsAttribute-driven layout Shneiderman, 2006 Semantic substrates

  43. Graphs Conclusion • Trees: • Indentation • Simple, effective for small trees • Node link and layered • Looks good but needs exponential space • Enclosure (treemaps) • Good for size related tasks but suffer in structure related tasks • Graphs: • Node link • Familiar, but problematic for dense graphs • Adjacency matrices • Abstract, hard to follow paths • Attribute-driven • Not always possible • No single “best” solution – a design problem

  44. Web Pages and VideosGraphs • vizster, social network, Facebook: http://vis.stanford.edu/jheer/projects/vizster/, (download *.wmv) • NY Times diseases: http://www.nytimes.com/interactive/2008/05/05/science/20080506_DISEASE.html • Force directed layout protovis: http://mbostock.github.io/protovis/ex/force.html • Protovis: http://vis.stanford.edu/protovis/ex/force.html • D3 (cf collapsible force directed): https://github.com/mbostock/d3/wiki/Gallery • Magnets and graphs: http://www.youtube.com/watch?v=K4GOxJywB-U • Force directed layout examples • 23 second example: http://www.youtube.com/watch?v=AYrkWSDkfLM • 60 second example: http://www.youtube.com/watch?v=QlXRapQW4q0 • Visual complexity: http://www.visualcomplexity.com/vc/

  45. End .

More Related