VisDB Visual Data Mining and Database Exploration

1
VisDB Visual Data Mining and Database Exploration

• Ming Luo
• Rui Yang

2
Characteristics

• Five Visualization Techniques
• Graphic Query Interface
• Graphic query result presentation
• Easy to slightly adjust the selection condition
  in a query

3
(No Transcript)
4
Graphic Query Interface
5
Basic Visualization Technique
Spiral arrangement and Coloration
Coloration Yellow is exactly matching. Black are
no relevance. Yellow, green, blue, red, brown,
black.
6
Calculating the relevance factors

• sp
• Combined Distance ? Wj dij in case of
  AND
• j1
• sp
• Combined Distance ? dijWj in case of OR
• j1
• Relevance factor is the inverse of distance value

7
Arrangement for Displaying five-dimensional data
8
2D Arrangement Displaying
9
Grouping Technique
10
Example 1

• EX.1 1,000 eight-dimensional data items with
  five clusters

11
Example 2

• EX.2 7,000 eight-dimensional data items with
  five clusters

12
Example 3

• EX.3 100,000 five dimensional data items
  containing five clusters

13
Thank you !

• Questions and discussion.