TreeMaps: A SpaceFilling Approach to the Visualization of Hierarchical Information Structures

1
Tree-Maps A Space-Filling Approach to the
Visualization of Hierarchical Information
Structures

• Brian Johnson
• Ben Shneiderman
• (HCIL TR 91-06)

Steve Betten February 14, 2001
2
Outline of Paper

• Introduction
• Comparison of methods
• Example directory tree
• Treemap method
• Algorithms
• Coping with size
• Future research directions

3
Introduction

• Motivation large hierarchical data
• Methods
• Spatial versus textual methods
• Interactive versus static methods
• Objectives
• Efficient space utilization, interactivity,
  comprehension, aesthetics
• Hierarchical data structure and content
• Treemap emphasis on structure and leaves

4
Comparison of Methods

• Motivating example
• Viewing large file system hierarchies
• Existing methods
• Listings, outlines, tree diagrams
• General problems
• Navigational difficulty
• Hidden content information
• Text display of content information
• Treemap

5
Comparison Existing Methods

• Listings (e.g. Unix ls, DOS dir)
• Good detailed content information
• Bad navigation difficulty with explicit paths or
  manual traversal
• Outlines (e.g., Unix du, Windows explorer)
• Good display of both structure and content
• Bad navigation difficulty only a few lines of
  nodes show at a time
• Listing and Outlines
• Bad required display space is linearly
  proportional to number of nodes

6
Comparison Existing Methods (Continued)

• Tree diagrams
• Good
• Effectiveness and pleasantness for small
  hierarchies
• Bad
• Inefficient space utilization
• Connections between nodes
• 50 of display is background
• Only partial success of zooming and panning
• Lack of content information in large hierarchies
• Cluttering from text
• For visual cues, insufficient size of nodes

7
Comparison Treemap

• Good
• Efficient utilization of display area
• Implicit display of structure
• No need to draw separate internal nodes
• More space for leaves and visual content cues
• Overview of entire hierarchy
• Rapid movement to any node
• Preservation of context
• Required display space is proportional to square
  root of number of nodes

8
Example Directory Tree

• Problems with existing methods
• Outline (Windows explorer) too many lines
• Tree diagram (Open Windows File Manager) wasted
  space
• Venn diagram wasted space
• Treemap (nested and non-nested)
• Easy identification of largest files on entire
  file system
• Easy identification of application, system, text,
  picture, and archive files

9
Treemap Method

• Structural information partitioning
• Weight (degree of interest) for each node
• Properties
• Node overlap only with ancestors or descendents
• Node display area proportional to weight
• Node weight ? sum of children's weights
• Structural information
• Implicitness via slice-and-dice
• Explicitness via additional nesting

10
Treemap Method (Continued)

• Content information
• Visual cues color, texture, blinking
• Popup windows that display content
• Auditory cues that precede popup windows
• User control of properties that decreases
  on-screen complexity

11
Algorithms

• Drawing of treemap
• Tracking of cursor movement
• Interactive display of node details

12
Coping with Size

• Average case analysis pixels per file
• Problem of small nodes not displaying
• Possible solutions magnification, zooming
• Insignificance (can usually ignore them)

13
Future Research Directions

• Alternative partitioning methods
• Top-down
• Visual cues for numeric and non-numeric content
  information
• Dynamic views (animation over time)
• Stock portfolio
• Node operations
• Zooming, marking, selecting, searching

14
Favorite Sentence

• Beard Users are never lost because they can see
  the complete information space.
• Expression of primary goal of treemap
• Use of 2D graphics and implicit internal nodes
• Display of entire hierarchy at once (structure)
• Significant visual cues for most nodes (content)
• Solution to previous navigation and content
  problems

15
Contributions

• Demonstration of application and effectiveness of
  treemap
• Provision of worthwhile direction for future
  research
• Alternative partitioning methods
• Applications stock (Smartmoney)
• Propagation of treemap concept
• Widely cited paper from 1991 IEEE Visualization
  conference

16
Critique

• Good
• Definition of problem domain
• Comparison to existing methods
• Concrete and relevant example
• Algorithms
• Bad
• Lack of emphasis on the difficulty of comparing
  rectangles with different aspect ratios
• Future research

17
Treemap Developments

• Academia
• 1992 stock portfolio visualization
• 1994 decision making
• 1994 satellite management
• 1999 cushion treemap squarified treemap
• 2000 parameterized rectangles
• 2000 TreeMap2000
• 2000 pivot by size pivot by position

• Industry
• DiskMapper
• SeeDiff software code viewer
• Storyspace hypertext authoring system
• Tcl/Tk widget
• Smartmoney
• PeopleMap
• Peets Coffee

18
Link Recommendations

• UMD HCIL history of treemap
• www.cs.umd.edu/hcil/treemaps
• TreeMap2000
• www.cs.umd.edu/hcil/treemaps/treemap2000
• Demo comparison of five treemap partitioning
  algorithms
• www.columbia.edu/mmw111/treemap/layout.html