Jaro similarity: Difference between revisions
Add C# implementation
mNo edit summary |
(Add C# implementation) |
||
| (3 intermediate revisions by 3 users not shown) | |||
Line 739:
0.8962963 'JELLYFISH' 'SMELLYFISH'
</pre>
=={{header|BBC BASIC}}==
{{works with|BBC BASIC for Windows}}
<syntaxhighlight lang="bbcbasic"> PRINT "Jaro similarity between the two strings:"
PROCDescribeJaro("MARTHA", "MARHTA")
PROCDescribeJaro("DIXON", "DICKSONX")
PROCDescribeJaro("JELLYFISH", "SMELLYFISH")
PROCDescribeJaro("DWAYNE", "DUANE")
PROCDescribeJaro("a", "b")
PROCDescribeJaro("", "")
END
DEF FNMax(a, b)=(a + b + ABS(a - b)) / 2
DEF FNMin(a, b)=(a + b - ABS(a - b)) / 2
DEF PROCDescribeJaro(word1$, word2$)
LOCAL d%, i%, j%, k%, l1%, l2%, m%, t%
PRINT " '" word1$ "' and '" word2$ "'" TAB(30) "= ";
IF word1$ == "" IF word2$ == "" PRINT;1 : ENDPROC
l1%=LENword1$
l2%=LENword2$
IF l1% < l2% SWAP l1%, l2% SWAP word1$, word2$
d%=l1% / 2 - 1
j%=1
FOR i%=1 TO l2%
IF MID$(word2$, i%, 1) == MID$(word1$, j%, 1) THEN
m%+=1
MID$(word1$, j%)=" "
ELSE
FOR k%=FNMax(1, i% - d%) TO FNMin(l1%, i% + d%)
IF MID$(word2$, i%, 1) == MID$(word1$, k%, 1) THEN
t%+=1
m%+=1
MID$(word1$, k%)=" "
IF k% > j% j%=k%
ENDIF
NEXT
ENDIF
j%+=1
NEXT
IF m% == 0 THEN
PRINT;0
ELSE
PRINT;(m% / l2% + m% / l1% + ((m% - (t% >> 1)) / m%)) / 3
ENDIF
ENDPROC</syntaxhighlight>
{{out}}
<pre>Jaro similarity between the two strings:
'MARTHA' and 'MARHTA' = 0.944444444
'DIXON' and 'DICKSONX' = 0.766666667
'JELLYFISH' and 'SMELLYFISH' = 0.896296296
'DWAYNE' and 'DUANE' = 0.822222222
'a' and 'b' = 0
'' and '' = 1</pre>
=={{header|C}}==
Line 836 ⟶ 892:
0.766667
0.896296
</pre>
=={{header|C#}}==
{{trans|Java}}
<syntaxhighlight lang="C#">
using System;
public class JaroDistance {
public static double Jaro(string s, string t) {
int s_len = s.Length;
int t_len = t.Length;
if (s_len == 0 && t_len == 0) return 1;
int match_distance = Math.Max(s_len, t_len) / 2 - 1;
bool[] s_matches = new bool[s_len];
bool[] t_matches = new bool[t_len];
int matches = 0;
int transpositions = 0;
for (int i = 0; i < s_len; i++) {
int start = Math.Max(0, i - match_distance);
int end = Math.Min(i + match_distance + 1, t_len);
for (int j = start; j < end; j++) {
if (t_matches[j]) continue;
if (s[i] != t[j]) continue;
s_matches[i] = true;
t_matches[j] = true;
matches++;
break;
}
}
if (matches == 0) return 0;
int k = 0;
for (int i = 0; i < s_len; i++) {
if (!s_matches[i]) continue;
while (!t_matches[k]) k++;
if (s[i] != t[k]) transpositions++;
k++;
}
return (((double)matches / s_len) +
((double)matches / t_len) +
(((double)matches - transpositions / 2.0) / matches)) / 3.0;
}
public static void Main(string[] args) {
Console.WriteLine(Jaro("MARTHA", "MARHTA"));
Console.WriteLine(Jaro("DIXON", "DICKSONX"));
Console.WriteLine(Jaro("JELLYFISH", "SMELLYFISH"));
}
}
</syntaxhighlight>
{{out}}
<pre>
0.944444444444445
0.766666666666667
0.896296296296296
</pre>
Line 2,832 ⟶ 2,952:
return 1 if $s eq $t;
my $
my $
my $
my ($matches, @
for ^@s -> $i {
my $start = 0 max $i - $
my $end = $i + $
for $start .. $end -> $j {
next if @
(@
$matches++ and last;
}
Line 2,851 ⟶ 2,971:
my ($k, $transpositions) = (0, 0);
for ^@s -> $i {
next unless @
$k++ until @
$transpositions++ if @s[$i] ne @t[$k];
$k++;
}
( $matches/$
}
say jaro(.key, .value).fmt: '%.3f' for
'MARTHA' => 'MARHTA', 'DIXON' => 'DICKSONX', 'JELLYFISH' => 'SMELLYFISH',
'I repeat myself' => 'I repeat myself', '' => '';
</syntaxhighlight>
{{out}}
<pre>0.944
Line 3,520 ⟶ 3,641:
{{trans|Go}}
{{libheader|Wren-fmt}}
<syntaxhighlight lang="
var jaro = Fn.new { |s1, s2|
| |||