Week14 Chap7 String Algorithms

Uploaded by

duonganhquanhsgs20

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

2 views

Week14 Chap7 String Algorithms

Uploaded by

duonganhquanhsgs20

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 13

APPLIED ALGORITHMS

CONTENTS

• Boyer Moore algorithm

• Rabin Karp algorithm
• KMP algorithm
APPLIED ALGORITHMS
STRING PROCESSING ALGORITHM

3 4
BOYER MOORE ALGORITHM BOYER MOORE ALGORITHM

a b a c b a c a c b a c a b a c b a c a c b a c

a c b a a c b a

5 6

BOYER MOORE ALGORITHM BOYER MOORE ALGORITHM

a b a c b a c a c b a c a b a c b a c a c b a c

a c b a a c b a

7 8
BOYER MOORE ALGORITHM BOYER MOORE ALGORITHM

a b a c b a c a c b a c a b a c b a c a c b a c

a c b a a c b a
Bad character c
• Slide the sample string P from left to right • Slide the sample string P from left to right
• Match: right to left j = 4, unmatch position • Match: right to left
• Use preprocessing information to skip as many • Use preprocessing information to skip as many
characters as possible characters as possible a c b a
• Preprocessing the sample string P • Preprocessing the sample string P
• Last[x]: The rightmost position that appears the • Last[x]: The rightmost position that appears the
letter x in P letter x in P
• When a mismatch occurs with the bad character x • When a mismatch occurs with the bad character x
(a character of T), P is slid to the right (a character of T), P is slid to the right
max{j - Last[x], 1} positions max{j - Last[x], 1} positions
where j is the current index (mismatch occurring) where j is the current index (mismatch occurring)
on P when matching characters from right to left on P when matching characters from right to left

9 10

BOYER MOORE ALGORITHM BOYER MOORE ALGORITHM

a b a c b a c a c b a c a b a c b a c a c b a c

a c b a a c b a

• Slide the sample string P from left to right • Slide the sample string P from left to right
• Match: right to left • Match: right to left
• Use preprocessing information to skip as many • Use preprocessing information to skip as many
characters as possible a c b a characters as possible a c b a
• Preprocessing the sample string P • Preprocessing the sample string P
• Last[x]: The rightmost position that appears the • Last[x]: The rightmost position that appears the
a c b a a c b a
letter x in P letter x in P
• When a mismatch occurs with the bad character x • When a mismatch occurs with the bad character x
(a character of T), P is slid to the right (a character of T), P is slid to the right a c b a
max{j - Last[x], 1} positions max{j - Last[x], 1} positions
where j is the current index (mismatch occurring) where j is the current index (mismatch occurring)
on P when matching characters from right to left on P when matching characters from right to left

11 12
BOYER MOORE ALGORITHM BOYER MOORE ALGORITHM

a b a c b a c a c b a c a b a c b a c a c b a c

a c b a a c b a

• Slide the sample string P from left to right • Slide the sample string P from left to right
• Match: right to left • Match: right to left
• Use preprocessing information to skip as many • Use preprocessing information to skip as many
characters as possible a c b a characters as possible a c b a
• Preprocessing the sample string P • Preprocessing the sample string P
• Last[x]: The rightmost position that appears the • Last[x]: The rightmost position that appears the
a c b a a c b a
letter x in P letter x in P
• When a mismatch occurs with the bad character x • When a mismatch occurs with the bad character x
(a character of T), P is slid to the right a c b a (a character of T), P is slid to the right a c b a
max{j - Last[x], 1} positions max{j - Last[x], 1} positions
where j is the current index (mismatch occurring) a c b a where j is the current index (mismatch occurring) a c b a
on P when matching characters from right to left on P when matching characters from right to left
a c b a
13 14

BOYER MOORE ALGORITHM CONTENTS

computeLast(p){ boyerMoore(P, T){

for c = 0 to 255 do last[c] = 0; computeLast(P);
• Boyer Moore algorithm
k = p.length();
for i = k-1 downto i >= 0 do {
s = 0; cnt = 0;
N = T.length(); M = P.length();
• Rabin Karp algorithm
if last[p[i]] = 0 then last[p[i]] = i; while s <= N-M do { • KMP algorithm
} j = M-1;
} while j >= 0 && T[j+s] = P[j] do
j = j -1;
if j == -1 then {
cnt++; s = s + 1;
}else{
k = last[T[j+s]];
s = s + (j - k > 1 ? j - k : 1);
}
}
return cnt;
}

15 16
RABIN KARP ALGORITHM RABIN KARP ALGORITHM

• The Rabin-Karp algorithm converts the sample strings to non-negative integers • Disadvantage
• Each letter in the alphabet is represented by a non-negative integer less than d • When M is large, converting strings to numbers takes considerable time,
• Convert the string P[1..M] to a positive integer • Can cause overflow for the basic data types of the programming language

p = P[1]*dM-1 + P[2]*dM-2 + . . . + P[M]*d0 • Solution: perform division by Q and get the remainder value
• Match patterns by comparing 2 corresponding code values: • When the 2 remainders are different, it means 2 different numeric values and 2 corresponding
• If the two codes are different, the two corresponding strings are different strings are also different
• If the two codes are equal, we proceed to match each character • When the two remainders are equal, match each character in the traditional way
• Use the Horner scheme to increase the speed of calculating the encoding of substrings in T
• With sliding position s, convert the substring T[s+1 .. s+M] to number:

Ts = T[s+1]dM-1 + T[s+2]dM-2 + . . . + T[s+M]*d0

• With sliding position s+1, Ts+1 can be efficiently calculated based on Ts (previously calculated):
Ts+1 = (Ts - T[s+1]*dM-1)*d + T[s+M+1]

17 18

RABIN KARP ALGORITHM CONTENTS

hashCode(p){ rabinKarp(P, T){

c = 0; cnt = 0; N = T.length(); M = P.length(); • Boyer Moore algorithm
for i = 0 to p.length()-1 do {
c = c*256 + p[i];
e = dM-1;
codeP = hashCode(P); codeT = hashCode(T,0,M-1);
• Rabin Karp algorithm
c = c%Q; for s = 0 to N-M do {
• KMP algorithm
} if(codeP = codeT){
return c; ok = true;
} for j = 0 to M-1 do if P[j] != T[j + s] then {
hashCode(s, start, end){ ok = false; break;
c = 0; }
for i = start to end do { if ok then cnt++;
c = c*256 + s[i]; }
c = c%Q; t = T[s]*e; t = t %Q; t = (codeT - t)%Q;
} codeT = (t*d + T[s+M])%Q;
return c; }
} return cnt;
}

19 20
KMP ALGORITHM KMP ALGORITHM

a b a c b a c a c b a c • Preprocessing:
• [q]: length of the longest prefix which is also the
strict suffix of the string P[1..q]
a c b a

• Slide the sample string P from left to right 1 2 3 4 5 6 7 8

• Match: left to right P a b a b a b c a
• Use preprocessing information to skip as many
characters as possible  0 0 1 2 3 4 0 1

21 22

KMP ALGORITHM KMP ALGORITHM

• Preprocessing: • Slide the sample string P from left to right over T

• [q]: length of the longest prefix which is also the
strict suffix of the string P[1..q] kmp(P, T){
q = 0;
computePI(P){
for i = 1..N do {
1 2 3 4 5 6 7 8 pi[1] = 0;
while q > 0 && P[q+1] != T[i]
P a b a b a b c a k = 0;
q = pi[q];
for q = 2  M do {
 0 0 1 2 3 4 0 1 while(k > 0 && P[k+1] != P[q])
if P[q+1] = T[i]
q = q + 1;
k = pi[k];
if(q = M){
if P[k+1] = P[q] then
output(i-M+1);
k = k + 1;
q = pi[q];
pi[q] = k;
}
}
}
}
}

23 24
KMP ALGORITHM KMP ALGORITHM