Tree Structures (Hierarchical Information) cs5764: Information Visualization Chris North.
-
Upload
shanon-bradford -
Category
Documents
-
view
219 -
download
3
Transcript of Tree Structures (Hierarchical Information) cs5764: Information Visualization Chris North.
Tree Structures(Hierarchical Information)
cs5764: Information Visualization
Chris North
Where are we?
• Multi-D• 1D• 2D• Trees• Graphs• 3D• Document collections
• Design Principles• Empirical Evaluation• Visual Overviews
Trees (Hierarchies)
• What is a tree?• DAG, one parent per node
• Items + structure (nodes + associations)
• In table model?• Add parent pointer attribute
• 1:M
Examples
• File system
• menus
• org charts
• Family tree
• classification/taxonomy
• Table of contents
• data structures
• …
Tasks
• Multi-D tasks, plus structure-based tasks:
• Find descendants, ancestors, siblings, cousins
• Overall structure, height, breadth, dense/sparse areas
• …
Tree Properties
• Structure vs. attributes• Attributes only (multi-dimensional viz)
• Structure only (1 attribute, e.g. name)
• Structure + attributes
• Branching factor
• Fixed level, categorical
Tree Visualization
• Example: TreeView
• Why is tree visualization hard?• Structure AND items
• Structure harder, consumes more space
• Data size grows very quickly (exponential)» #nodes = bheight
2 Approaches
• Connection (node & link)
• outliner
• Containment (node in node)
• Venn diagram
A
CB
A
B C
Connection (node & link)
TreeView
• Good for directed search tasks
• subtree filtering (+/-)
• Not good for learning structure
• No attributes
• Apx 50 items visible
• Lose path to root for deep nodes
• Scroll bar!
Mac FinderBranching factor:
Small
large
Hyperbolic Trees
• Rao, “Hyperbolic Tree”•
• http://startree.inxight.com/
• Xerox PARC
• Inxight
• Focus+context
Cone Trees
• Robertson, “ConeTrees”•
• Xerox PARC
• 3D for focus+context
PDQ Trees
• Overview+Detail of 2D tree layout
• Dynamic Queries on each level for pruning
PDQ Trees
Disk Tree
• Ed Chi, Xerox PARC
• Overview:Reduced visual representation
WebTOC• Website map: TreeView + size attributes• http://www.cs.umd.edu/projects/hcil/webtoc/fhcil.html
FSN
• SGI file system navigator
• Jurassic Park
• Zooming?
Ugh!
Containment (node in node)
2 Approaches
• Connection (node & link)
• Outliner
• Containment (node in node)
• Venn diagram
• Structure vs. attributes• Attributes only (multi-dimensional viz)
• Structure only (1 attribute, e.g. name)
• Structure + attributes
A
CB
A
B C
Pyramids
3D Containment
Treemaps
• Shneiderman, “Treemaps”•
• http://www.cs.umd.edu/hcil/treemap3/
• Maryland
• zooming
Treemap Algorithm
• Calculate node sizes:• Recurse to children
• node size = sum children sizes
• Draw Treemap (node, space, direction)• Draw node rectangle in space
• Alternate direction (slice or dice)
• For each child:– Calculate child space as % of node space using size and direction
– Draw Treemap (child, child space, direction)
Squarified Treemaps• Wattenberg
• Van Wijk
• http://www.research.microsoft.com/~masmith/all_map.jpg
Cushion Treemaps• Van Wijk • http://www.win.tue.nl/sequoiaview/
Dynamic Query Treemaps• http://www.cs.umd.edu/hcil/treemap3/
Treemaps on the Web• Map of the Market: http://www.smartmoney.com/marketmap/
• People Map: http://www.truepeers.com/
• Coffee Map: http://www.peets.com/tast/11/coffee_selector.asp
DiskMapper
• http://www.miclog.com/dmdesc.htm
Sunburst
• Stasko, GaTech
• Radial layout
• Animated zooming
Sunburst (vs. Treemap)
• + Faster learning time: like pie chart• + Details outward, instead of inward• + Focus+context instead of zooming
• - Not space filling• - More space used by non-leaves• - Less scalability?
• All leaves on 1-D space, perimeter• Treemap: 2-D space for leaves
Misc.
CHEOPS
• Beaudoin, “Cheops”•
• http://www.crim.ca/hci/cheops/index1.html
• http://tecfa.unige.ch/~schneide/cheops/lite1.html
The Original Fisheye View
• George Furnas, 1981 (pg 311)
• Large information space
• User controlled focus point
• How to render items?• Normal View: just pick items nearby
• Fisheye View: pick items based on degree of interest
• Degree of Interest = function of distance from f and a priori importance
• DOI(x) = -dist(x,f) + imp(x)x
f
Example: Tree structure
• Distance = # links between f and x
• Importance = level of x in tree
Distance:
I A a i ii b i ii B a i ii b i ii
Importance:
I A a i ii b i ii B a i ii b i ii
DOI:
I A a i ii b i ii B a i ii b i ii
f
Challenges
• Multiple foci
• George Robertson, Microsoft Research
Polyarchies
• multiple inter-twined trees
• Visual pivot• George Robertson, Microsoft Research
Nifty App of the Day
• SAS JMP
Summary
• Hyperbolic <1000
• TreeMap <3000, attributes, collective
• Cheops = scale up