DIRECT ACCESS FILES

1
DIRECT ACCESS FILES

• instructor ebru a. sezer
• esezer_at_ata.cs.hun.edu.tr

2
Purpose of using hashes

• Achieve O(1)
• For efficient evaluation of equality selections

3
Hashing

• A hash function is a black box that produce an
  address every time you drop in a key
• Hash function converts key K to address
• Note Now lets think address as record number
  in a file (or table)

4
Example

• Hash function Convert key in Ascii format and
  multiply first two character and use rightmost
  three digit
• H(Lowell) 4

5
Good Hash Function

• Generated record numbers should be uniformly and
  randomly disrubuted over the file (0 lt h(key)lt
  N)
• Small variation in the value of the key will
  cause large value of H(key)
• The hashing function must minimixe the creation
  of synonyms (minimize collisions).

6
Load Factor

• Load Factor Logical record number /
• (Block number Record per block)
  
• If address size increases
• Collision decreases
• LF decreases
• LF lt 0.6 is unacceptable
• Reduce collisions while keeping LF in acceptable
  level

7
Hashing Transformations-1

• Digit analysis
• Select determined place of digits from key
• Division method
• f(x) x mod n
• Radix transformation
• f(abc) a 112 b 11 c

8
Hashing Transformations-2

• Truncation Method
• f(k) k mod 10n
• Produce n digit address
• Midsquare Method
• f(k) Midy (k2)

9
Hashing Tranformations-3

• Shift Folding
• Key is partitioned from left to right in
  subnumbers
• Each subnumber contains digit in the rqd.address
• Subnumbers added, carry discarded
• Bounding Folding
• Same as shift folding
• First and last subnumbers are reverse

10
Overflow Managment Techniques

• Open Addressing
• Linear search
• Nonlinear search
• Chaining
• Extendible Hashing
• Linear Hashing

11
Open Addressing

• When collision occurs, the new record will be
  inserted in the first address space after the
  hashed address
• Linear Search
• With pump
• Without pump

12
Insert Algorithm-Linear Search

• Algorithm Insert
• home_address h(k)
• i home_address
• found False
• while ((location i is full) and (not found) ) do
  
• i (i1) mod (address space)
• if (i home_address) then found True
• if (found) then write (disk full)
• else insert record in location i

13
Search Length

• Average search length
• total search length / total number of records
• Total search length
• sum of each records search length

14
Open Addressing-Nonlinear Search

• (a) Rehashing
• Employs a second hashing function
• For overflow records, rehashing is used until
  free location is found or a full file is detected
• r(i) (i p) mod q

15
Insert Algorithm-Rehashing

• Algorithm Insert
• home_address h(k)
• i home_address
• found False
• while ((location i is full) and (not found) ) do
  
• i r(i)
• if (i home_address) then found True
• if (found) then write (disk full)
• else insert record in location i

16
Open Addressing-Nonlinear Search

• (b) Double Hashing
• Has two hashing function h1 and h2
• For overflow records, h2 is used until free
  location is found or a full file is detected
• r(i) (i h2(k)) mod q
• Purpose is to cancel similarties of to synonym
  for rehashing

17
(No Transcript)