Lecture 22 - SS-trees,SR-trees,P-trees Notes | EduRev

: Lecture 22 - SS-trees,SR-trees,P-trees Notes | EduRev

 Page 1


Objectives_template
file:///C|/Documents%20and%20Settings/iitkrana1/My%20Documents/Google%20Talk%20Received%20Files/ist_data/lecture22/22_1.htm[6/14/2012 3:42:29 PM]
 Module 5: Disk-based Index Structures
 Lecture 22: SS-trees,SR-trees,P-trees
 
                                           
The Lecture Contains:
SS-tree
Bounding spheres and bounding rectangles
SR-tree
Structure and algorithms
Performance
P-tree
Search performance and overlap
Effect of dimensionality on overlap
 
 
 
 
 
 
 
 
 
 
 
 
 
Page 2


Objectives_template
file:///C|/Documents%20and%20Settings/iitkrana1/My%20Documents/Google%20Talk%20Received%20Files/ist_data/lecture22/22_1.htm[6/14/2012 3:42:29 PM]
 Module 5: Disk-based Index Structures
 Lecture 22: SS-trees,SR-trees,P-trees
 
                                           
The Lecture Contains:
SS-tree
Bounding spheres and bounding rectangles
SR-tree
Structure and algorithms
Performance
P-tree
Search performance and overlap
Effect of dimensionality on overlap
 
 
 
 
 
 
 
 
 
 
 
 
 
Objectives_template
file:///C|/Documents%20and%20Settings/iitkrana1/My%20Documents/Google%20Talk%20Received%20Files/ist_data/lecture22/22_2.htm[6/14/2012 3:42:29 PM]
 Module 5: Disk-based Index Structures
 Lecture 22: SS-trees,SR-trees,P-trees
 
                                           
 
 
SS-tree
Uses bounding spheres instead of bounding rectangles
 
 
 
 
 
 
 
 
 
 
 
 
 
Page 3


Objectives_template
file:///C|/Documents%20and%20Settings/iitkrana1/My%20Documents/Google%20Talk%20Received%20Files/ist_data/lecture22/22_1.htm[6/14/2012 3:42:29 PM]
 Module 5: Disk-based Index Structures
 Lecture 22: SS-trees,SR-trees,P-trees
 
                                           
The Lecture Contains:
SS-tree
Bounding spheres and bounding rectangles
SR-tree
Structure and algorithms
Performance
P-tree
Search performance and overlap
Effect of dimensionality on overlap
 
 
 
 
 
 
 
 
 
 
 
 
 
Objectives_template
file:///C|/Documents%20and%20Settings/iitkrana1/My%20Documents/Google%20Talk%20Received%20Files/ist_data/lecture22/22_2.htm[6/14/2012 3:42:29 PM]
 Module 5: Disk-based Index Structures
 Lecture 22: SS-trees,SR-trees,P-trees
 
                                           
 
 
SS-tree
Uses bounding spheres instead of bounding rectangles
 
 
 
 
 
 
 
 
 
 
 
 
 
Objectives_template
file:///C|/Documents%20and%20Settings/iitkrana1/My%20Documents/Google%20Talk%20Received%20Files/ist_data/lecture22/22_3.htm[6/14/2012 3:42:29 PM]
 Module 5: Disk-based Index Structures
 Lecture 22: SS-trees,SR-trees,P-trees
 
                                           
 
 
SS-tree
Uses bounding spheres instead of bounding rectangles
Motivated by range and kNN queries which are hyper-spheres
Center of sphere is centroid of points
Maintains total number of points in the subtree
Higher fanout due to smaller storage requirements
Considerable volume overlap
Split axis is chosen based on variance
Better than R*-trees
 
 
 
 
 
 
 
 
 
 
 
 
 
Page 4


Objectives_template
file:///C|/Documents%20and%20Settings/iitkrana1/My%20Documents/Google%20Talk%20Received%20Files/ist_data/lecture22/22_1.htm[6/14/2012 3:42:29 PM]
 Module 5: Disk-based Index Structures
 Lecture 22: SS-trees,SR-trees,P-trees
 
                                           
The Lecture Contains:
SS-tree
Bounding spheres and bounding rectangles
SR-tree
Structure and algorithms
Performance
P-tree
Search performance and overlap
Effect of dimensionality on overlap
 
 
 
 
 
 
 
 
 
 
 
 
 
Objectives_template
file:///C|/Documents%20and%20Settings/iitkrana1/My%20Documents/Google%20Talk%20Received%20Files/ist_data/lecture22/22_2.htm[6/14/2012 3:42:29 PM]
 Module 5: Disk-based Index Structures
 Lecture 22: SS-trees,SR-trees,P-trees
 
                                           
 
 
SS-tree
Uses bounding spheres instead of bounding rectangles
 
 
 
 
 
 
 
 
 
 
 
 
 
Objectives_template
file:///C|/Documents%20and%20Settings/iitkrana1/My%20Documents/Google%20Talk%20Received%20Files/ist_data/lecture22/22_3.htm[6/14/2012 3:42:29 PM]
 Module 5: Disk-based Index Structures
 Lecture 22: SS-trees,SR-trees,P-trees
 
                                           
 
 
SS-tree
Uses bounding spheres instead of bounding rectangles
Motivated by range and kNN queries which are hyper-spheres
Center of sphere is centroid of points
Maintains total number of points in the subtree
Higher fanout due to smaller storage requirements
Considerable volume overlap
Split axis is chosen based on variance
Better than R*-trees
 
 
 
 
 
 
 
 
 
 
 
 
 
Objectives_template
file:///C|/Documents%20and%20Settings/iitkrana1/My%20Documents/Google%20Talk%20Received%20Files/ist_data/lecture22/22_4.htm[6/14/2012 3:42:30 PM]
 Module 5: Disk-based Index Structures
 Lecture 22: SS-trees,SR-trees,P-trees
 
                                           
 
 
Bounding spheres and bounding rectangles
SS-tree has regions with shorter diameters
R*-tree has regions with smaller volumes
How?
 
 
 
 
 
 
 
 
 
 
 
 
 
Page 5


Objectives_template
file:///C|/Documents%20and%20Settings/iitkrana1/My%20Documents/Google%20Talk%20Received%20Files/ist_data/lecture22/22_1.htm[6/14/2012 3:42:29 PM]
 Module 5: Disk-based Index Structures
 Lecture 22: SS-trees,SR-trees,P-trees
 
                                           
The Lecture Contains:
SS-tree
Bounding spheres and bounding rectangles
SR-tree
Structure and algorithms
Performance
P-tree
Search performance and overlap
Effect of dimensionality on overlap
 
 
 
 
 
 
 
 
 
 
 
 
 
Objectives_template
file:///C|/Documents%20and%20Settings/iitkrana1/My%20Documents/Google%20Talk%20Received%20Files/ist_data/lecture22/22_2.htm[6/14/2012 3:42:29 PM]
 Module 5: Disk-based Index Structures
 Lecture 22: SS-trees,SR-trees,P-trees
 
                                           
 
 
SS-tree
Uses bounding spheres instead of bounding rectangles
 
 
 
 
 
 
 
 
 
 
 
 
 
Objectives_template
file:///C|/Documents%20and%20Settings/iitkrana1/My%20Documents/Google%20Talk%20Received%20Files/ist_data/lecture22/22_3.htm[6/14/2012 3:42:29 PM]
 Module 5: Disk-based Index Structures
 Lecture 22: SS-trees,SR-trees,P-trees
 
                                           
 
 
SS-tree
Uses bounding spheres instead of bounding rectangles
Motivated by range and kNN queries which are hyper-spheres
Center of sphere is centroid of points
Maintains total number of points in the subtree
Higher fanout due to smaller storage requirements
Considerable volume overlap
Split axis is chosen based on variance
Better than R*-trees
 
 
 
 
 
 
 
 
 
 
 
 
 
Objectives_template
file:///C|/Documents%20and%20Settings/iitkrana1/My%20Documents/Google%20Talk%20Received%20Files/ist_data/lecture22/22_4.htm[6/14/2012 3:42:30 PM]
 Module 5: Disk-based Index Structures
 Lecture 22: SS-trees,SR-trees,P-trees
 
                                           
 
 
Bounding spheres and bounding rectangles
SS-tree has regions with shorter diameters
R*-tree has regions with smaller volumes
How?
 
 
 
 
 
 
 
 
 
 
 
 
 
Objectives_template
file:///C|/Documents%20and%20Settings/iitkrana1/My%20Documents/Google%20Talk%20Received%20Files/ist_data/lecture22/22_5.htm[6/14/2012 3:42:30 PM]
 Module 5: Disk-based Index Structures
 Lecture 22: SS-trees,SR-trees,P-trees
 
                                           
 
 
Bounding spheres and bounding rectangles
SS-tree has regions with shorter diameters
R*-tree has regions with smaller volumes
How?
Rectangles in -dimensions
 
Diameter: Between 0 and 
 
Volume: Between 0 and 1
Spheres in -dimensions
 
Diameter: Between 0 and 
 
Volume: Between 0 and 
 
 
 
 
 
 
 
 
 
 
 
 
 
Read More
Offer running on EduRev: Apply code STAYHOME200 to get INR 200 off on our premium plan EduRev Infinity!