Module 3: Hashing
  Lecture 11: Locality Sensitive Hashing and Grid File
 

                                            

 

 

Grid file
  • Multi-dimensional hashing
  • Goal: To retrieve records with at most two disk accesses
  • Two parts:
 
  • Grid directory
 
  • Buckets (or pages)
  • Buckets hold data records
  • Grid directory maintains grid cells that points to buckets
  • Two parts:
 
  • -dimensional array: One entry for each grid cell
 
  • 1-dimensional linear scales: Defines partitions for each dimension
  • Arrangement of records inside a bucket can be anything