Visualization Schemas for Flexible Information Visualization

1
Visualization Schemas for Flexible Information
Visualization

• Chris North, Nathan Conklin, Varun Saini
• Proceedings of IEEE Symposium on InforVis02
• Presented by
• Mei Huang, Chunyuan Liao
• Apr. 21,2005

2
Outline

• Relational Data Schema
• Motivation
• Related Work
• Snap-Together
• Datacompass
• Summary Remarks

3
Relational Data Schema

• Structural description of data sets
• Primitives attributes, tuples and relations

4
Motivation

• Relational data schema enables flexible database
  design
• No corresponding flexible ways to construct
  effective UI and visualization
• -- unique data schema ? unique visualization/coord
  ination
• -- database keeps changing
• -- different views for same data

5
Mismatch in design capabilities
Relational Databases Traditional Visualization
Design Goal Data design Visualization design
Design Method Data schema Program code
Designer Data owner Programmer only
Design Change Rapid, dynamic Slow, static
Adaptability Flexible Brittle
6
Related Work

• Single relation visualization
• APT
• Sage/SageBrush
• DEVise
• Multiple relation visualization
• Visage
• DataSplash/Tioga-2
• Rivet/Polaris
• Sieve

7
DEVise
http//www.cs.wisc.edu/devise/devise/quick_intro/
index.html
8
Visage
www-2.cs.cmu.edu/sage/visage.html
9
DataSplash/Tioga-2
http//datasplash.cs.berkeley.edu/tour_quick.html
10
Polaris
http//graphics.stanford.edu/projects/polaris/
11
Snap-Together -- Overview
A strong analogy between relational database
concepts and Snap visualization concepts enables
a matching level of design capability.
Demo
12
Snap-Together -- Theory(1)

• Snap Visualization Model
• Multiple views/Components
• Schema primitives
• Data-centric coordination and Joins

13
Snap-Together -- Theory(2)

• Self Join
• eg TreeView ? URLs ? Table view
• Single Join
• eg TreeView ? URLs ?? HitCounts ?
  Scatterplot
• Compound Join
• eg TreeView ? URLs ??Hits ?? Referrers ?
  TableView
• Multiple alternative join
• eg
• TreeView ? URLs ??Hits ?? Referrers ?
  TableView
• TreeView ? URLs ??Links ?? Referrers ?
  TableView

url_id
url_id
refer_id
url_id
refer_id
refer_id
url_id
14
Snap-Together -- User Interface

• Visualization Schemas
• -- represented as a graph and support direct
  manipulation
• Nodes
• -- represent instantiated visualization
  components
• Edges
• -- represent coordinations between visualizations

15
Snap-Together -- System Architecture
Theory UI Architecture
Coordinated Muti-views
Visualization Model -Visualization -Coordination
Relational Model - Relation - Association
Data Source
Coordination Manager
Coordination Graph
Visualization Schema
Database Schema
Database Manager
Relational Database
Relational Database
16
Snap server

• Event-based coordination
• Send -gt Translate -gt Receive -gt
  selection/navigate
• Extendable architecture

17
DataCompass

• For novice users or very complex database schemas
• One-step construction
• Interchangeable with visualization schema
• Bottom-up approach vs. Top down approach

18
Summary -- Snaps three perspectives

• Theory multi-view visualization, coordinating
  between data design and visualization design
• UI diagrammatic UI to enable rapid customization
  of visualization without programming
• System Architecture web-based component
  architecture to support run-time integration of
  diverse data sources and visualization tools, and
  dissemination of custom visualization as web
  pages.

19
Remarks

• Merits
• A cohesive and extensible architecture for
  coordinating visualization components
• Flexible and easy user interface, no programming
  needed
• Shortcoming
• No support for visual query
• No integration between query and visualization
  schema
• Limited support for coordinated data navigation (
  pan, zoom )

20
Thanks!