Slides from 2007 talk at the IEEE Conference on Information Visualization

1. VisLink: Revealing Relationships
Amongst Visualizations
C H R I S TO P H E R C O L L I N S
&
S H E E L AG H C A R P E N DA L E
ccollins@cs.utoronto.ca, sheelagh@ucalgary.ca
IEEE Information Visualization 2007

3. (Collins, 2007)

4. (Heer, 2006 [prefuse]) & (Fry, 2004)

5. Understanding Multiple Relations
 What is the relationship…
 across different views of the same data?
 across different relations in the same dataset?
 across multiple relations and datasets?

6. VisLink

7. VisLink Overview
 Any number of 2D
visualizations, each on its
own plane in 3D space
 Adjacent planes connected
by bundled edges
 Shortcuts and constrained
widgets aid usability
 Enables powerful
inter-visualization queries

8. Formalizing Multiple Relations Visualizations
Dataset Relation Visualization
Conference Attendee Data Professor / Student Node-link social
network graph
Formalism for Multiple Relationship Visualization Comparison

9. Formalizing Multiple Relations Visualizations
Dataset Relation Visualization
DA
Formalism for Multiple Relationship Visualization Comparison

10. Formalizing Multiple Relations Visualizations
Dataset Relation Visualization
DA R A ( DA )
Formalism for Multiple Relationship Visualization Comparison

11. Formalizing Multiple Relations Visualizations
Dataset Relation Visualization
DA R A ( DA ) Vis A  RA ( DA )
Formalism for Multiple Relationship Visualization Comparison

12. Formalizing Multiple Relations Visualizations
Relation
R A ( DA )
Dataset Visualization
DA Vis A  RA ( DA )
Formalism for Multiple Relationship Visualization Comparison

13. Formalizing Multiple Relations Visualizations
Relation
R A ( DA )
Dataset Visualization
Relation
DA RB ( DA ) Vis A  RA ( DA )
Formalism for Multiple Relationship Visualization Comparison

14. Formalizing Multiple Relations Visualizations
Visualization
Relation Vis A  RA ( DA )
R A ( DA )
Visualization
Dataset
Vis B  RA ( DA )
Relation
DA RB ( DA )
Formalism for Multiple Relationship Visualization Comparison

15. Formalizing Multiple Relations Visualizations
Visualization
Relation Vis A  RA ( DA )
R A ( DA )
Visualization
Dataset
Vis B  RA ( DA )
Relation
DA RB ( DA )
Visualization
VisC  RB ( DA )
Formalism for Multiple Relationship Visualization Comparison

16. Multiple Relation Visualizations
Individual Visualizations
Coordinated Views
Compound Graphs
Semantic Substrates
VisLink
Formalism for Multiple Relationship Visualization Comparison

17. Individual Visualizations
 Any datasets, relations, and visualizations
 Manually compare
 e.g. different charts in Excel
Formalism for Multiple Relationship Visualization Comparison

18. Coordinated Views
Formalism for Multiple Relationship Visualization Comparison

19. Coordinated Views
Vis A  RA ( DA )
Formalism for Multiple Relationship Visualization Comparison

20. Coordinated Views
Vis A  RA ( DA ) Vis A  RB ( DA )
Formalism for Multiple Relationship Visualization Comparison

21. Coordinated Views
Vis A  RA ( DA ) Vis A  RB ( DA )
 Any datasets, relations, and visualizations
 Interactive highlighting
 e.g., Snap-Together Visualization (North & Shneiderman, 2000)
Formalism for Multiple Relationship Visualization Comparison

22. Compound Graphs
Formalism for Multiple Relationship Visualization Comparison

23. Compound Graphs
Vis A  RA ( DA )
Formalism for Multiple Relationship Visualization Comparison

24. Compound Graphs
Vis A  RA ( DA )  RB ( DA )
Formalism for Multiple Relationship Visualization Comparison

25. Compound Graphs
Vis A  RA , RB ( DA )
 Secondary relation has no spatial rights
 e.g., Overlays on Treemaps (Fekete et al., 2003), ArcTrees
Use of the powerful spatial dimension
(Neumann et al., 2005), Hierarchical Edge Bundles (Holten, 2006)
to encode data relationships.
Formalism for Multiple Relationship Visualization Comparison

26. Semantic Substrates
Formalism for Multiple Relationship Visualization Comparison

27. Semantic Substrates
DA
Formalism for Multiple Relationship Visualization Comparison

28. Semantic Substrates
DA1
DA
Formalism for Multiple Relationship Visualization Comparison

29. Semantic Substrates
DA1
DA2
DA
Formalism for Multiple Relationship Visualization Comparison

30. Semantic Substrates
DA1
DA2
DA
…
DAn
Formalism for Multiple Relationship Visualization Comparison

31. Semantic Substrates
DA1 DA2
Formalism for Multiple Relationship Visualization Comparison

32. Semantic Substrates
Vis A  RA ( DA1 ) Vis A  RA ( DA2 )
Formalism for Multiple Relationship Visualization Comparison

33. Semantic Substrates
Vis A  RA ( DA1 ) Vis A  RA ( DA2 )
Vis A  RA ( DA )
Formalism for Multiple Relationship Visualization Comparison

34. Semantic Substrates
 Single visualization, single relation
 Semantically meaningful data subsets
 Spatial rights for all relations
(Shneiderman and Aris, 2006)
Formalism for Multiple Relationship Visualization Comparison

35. VisLink
Formalism for Multiple Relationship Visualization Comparison

36. VisLink
Vis A  RA ( DA )
Formalism for Multiple Relationship Visualization Comparison

37. VisLink
Vis A  RA ( DA ) Vis B  RB ( DA )
Formalism for Multiple Relationship Visualization Comparison

38. VisLink
Vis A  RA ( DA ) Vis B  RB ( DA )
Vis B  RA ( DB )
Formalism for Multiple Relationship Visualization Comparison

39. VisLink
Vis A  RA ( DA ) Vis B  RB ( DA )
Vis A B  T ( RA ( DA ), RB ( DA ))
Formalism for Multiple Relationship Visualization Comparison

40. VisLink
Vis A B  T ( RA ( DA ), RB ( DA ))
 Visualize second order relations between visualizations
 Across any datasets, relations, visualizations for which a
relation can be defined
 All component visualizations retain spatial rights
Formalism for Multiple Relationship Visualization Comparison

41. VisLink & Semantic Substrates
Vis A  RA ( DA1 ) Vis A  RA ( DA2 ) Vis A  RA ( DA ) Vis B  RB ( DA )
Vis A  RA ( DA ) Vis A B  T ( RA ( DA ), RB ( DA ))
Formalism for Multiple Relationship Visualization Comparison

42. VisLink & Semantic Substrates
Vis A  RA ( DA1 ) Vis A  RA ( DA2 ) Vis A  RA ( DA ) Vis B  RB ( DA )
Vis A  RA ( DA ) Vis A B  T ( RA ( DA ), RB ( DA ))
 Single visualization technique
 Semantic subsets of data provide added meaning
Formalism for Multiple Relationship Visualization Comparison

43. VisLink & Semantic Substrates
Vis A  RA ( DA1 ) Vis A  RA ( DA2 ) Vis A  RA ( DA ) Vis B  RB ( DA )
Vis A  RA ( DA ) Vis A B  T ( RA ( DA ), RB ( DA ))
 Any number of different relations and visualizations
 Second order relations revealed in inter-plane edges
Formalism for Multiple Relationship Visualization Comparison

44. Equivalency & Extension
Formalism for Multiple Relationship Visualization Comparison

45. VisLink
VisLink Visualization

46. VisLink Case Study: Lexical Data
WordNet IS-A hierarchy (RA)
? Similarity clustering (RB) using
using radial tree (VisA) force-directed layout (VisB)
VisLink Visualization

47. Edge Detail
 Bundled:
one-to-many edges
 Smooth:
Chaiken corner cutting
 Transparent:
bundles more opaque
 Directed:
orange-to-green
VisLink Visualization

48. Interaction With Component Visualizations
 Always equivalent to 2D:
 Planes are virtual displays
 Mouse events transformed and passed to underlying visualization
 Equivalent to 2D viewing mode
VisLink Visualization

49. Interplane Edges
VisLink Visualization

50. Zoom
VisLink Visualization

51. Filter
VisLink Visualization

52. Constrained Widget Interaction
VisLink Interaction

53. 3D Navigation
VisLink Interaction

54. Spreading Activation
 Nodes have a level of activation, indicated by
transparency of orange node background
 Full activation through:
 Selecting a node on a plane
 Node matches search query
 Activation propagates through interplane edges,
reflecting between planes with exponential drop-off
 Enables inter-visualization queries
 Edge transparency relative to source node activation
Spreading Activation

55. Inter-Plane Query Example
2: synonym sets
1: alphabetic clusters
No synonym information No alphabetic organization
Q: Synonyms in the alphabetic view?
Spreading Activation

56. Inter-Plane Query Example
1. Select a word on plane 1
2. Edges propagate to synonym sets on plane 2
3. Reflected edges propagate back, revealing
synonyms in alphabetic clusters
1: similarity clusters 2: synonym sets
Spreading Activation

57. Inter-Plane Query Example
1. Select a word on plane 1
2. Edges propagate to synonym sets on plane 2
3. Reflected edges propagate back, revealing
synonyms in alphabetic clusters
1: similarity clusters 2: synonym sets
Spreading Activation

58. Inter-Plane Query Example
1. Select a word on plane 1
2. Edges propagate to synonym sets on plane 2
3. Reflected edges propagate back, revealing
synonyms in alphabetic clusters
1: similarity clusters 2: synonym sets
Spreading Activation

59. Inter-Plane Query Example
1. Select a word on plane 1
2. Edges propagate to synonym sets on plane 2
3. Reflected edges propagate back, revealing
synonyms in alphabetic clusters
1: similarity clusters 2: synonym sets
Spreading Activation

60. Edge Reflection and Inter-Plane Queries
Spreading Activation

61. Linking Existing Visualizations
Case Study

62. Linking Existing Visualizations
Case Study

63. Implementation
 Prefuse visualization toolkit (Heer et al., 2005)
 Existing visualizations can be incorporated without changes
 Interplane edges defined by (plane, node) index pairs
 Java OpenGL
Case Study

64. Perceptual Considerations
 Not all layouts equal
 Colour interactions with edges and visualizations
 3D perspective bias
Conclusion

65. Future Work
 Application to additional analytic scenarios
 Investigation of 3D edge bundling, edge lenses
 Animation of spreading activation
 Evaluation against existing multiple view techniques
 Rich query language to filter visualization planes
Conclusion

66. Summary
 Formalism to describe multi-relation visualizations
 New way to reveal relationships amongst visualizations
 Reuse of the powerful spatial visual dimension
 Full 2D interactivity for constituent visualizations
 Techniques to simplify 3D navigation
 Visualization bridging through inter-representational
queries and spreading activation
Conclusion

67. VisLink: Revealing Relationships
Amongst Visualizations
ACKNOWLEDGEMENTS:
Gerald Penn, Petra Isenberg, Mark Hancock,
Tobias Isenberg, Uta Hinrichs, and Matthew Tobiasz,
and helpful reviewers for their advice.
Christopher Collins (ccollins@cs.utoronto.ca) & Sheelagh Carpendale (sheelagh@ucalgary.ca)