Module 5: Disk-based Index Structures
  Lecture 21: R-tree Variants
 

                                           

 

 

Splitting
  • Two parameters:
 
  • Margin: Margin(MBR(Group 1)) + Margin(MBR(Group 2))
 
  • Overlap: Volume(MBR(Group 1) MBR(Group 2))
  • Choose split axis
 
  • For each axis, sort entries by low values
 
  • Enumerate possible groupings or distributions
 
  • kth distribution: First entries and rest
 
  • Compute sum of margins along that axis of all distributions
 
  • Do the same by sorting high values of each axis
 
  • Choose axis with least sum of margin values
 
  • For chosen axis and low/high value, choose distribution with least overlap