Module 8: Miscellaneous Topics
  Lecture 37: Aggregation
 

                                            

 

 

Analysis
  • Correct
  • Any object not seen cannot be in top- list
  • Due to monotonicity property
 
  • in earlier example
  • Comparison with brute-force (BF) algorithm
  • Assume a disk page contains one object and its associated score
  • BF does random accesses and sequential accesses
  • Denote cost of random access by and sequential access by
  • In earlier example
 
  • Database access cost of BF is
 
  • Cost of FA is
  • FA does lesser sequential accesses but more random accesses
  • " Seen" list is long leading to many random accesses
  • Finding "full" objects may need deep probing into the lists