NAME
Algorithm::HowSimilar - Perl extension for quantifying similarites between things
SYNOPSIS
use Algorithm::HowSimilar qw(compare);
@res = compare( $str1, $str2, sub { s/\s+//g; [split //] } );
@res = compare( \@ary1, \@ary2 );
DESCRIPTION
This module leverages Algorithm::Diff to let you compare the degree of sameness of array or strings. It returns a result set that defines exactly how similar these things are.
METHODS
compare( ARG1, ARG2, OPTIONAL_CALLBACK )
You can call compare with either two strings compare( $str1, $str2 ):
my ( $av_similarity,
$sim_str1_to_str2,
$sim_str2_to_str1,
$matches,
$in_str1_but_not_str2,
$in_str2_but_not_str1
) = compare( 'this is a string-a', 'this is a string bbb' );
Note that the mathematical similarities of one string to another will be different unless the strings have the same length. The first result returned is the average similarity. Totally dissimilar strings will return 0. Identical strings will return 1. The degree of similarity therefore ranges from 0-1 and is reported as the biggest float your OS/Perl can manage.
You can also compare two array refs compare( \@ary1, \@ary2 ):
my ( $av_similarity,
$sim_ary1_to_ary2,
$sim_ary2_to_ary1,
$ref_ary_matches,
$ref_ary_in_ary1_but_not_ary2,
$ref_ary_in_ary2_but_not_ary1
) = compare( [ 1,2,3,4 ], [ 3,4,5,6,7 ] );
When called with two string you can specify an optional callback that changes the default tokenization of strings (a simple split on null) to whatever you need. The strings are passed to you callback in $_ and the sub is expected to return an array ref. So for example to ignore all whitespace you could:
@res = compare( 'this is a string',
'this is a string ',
sub { s/\s+//g; [split //] }
);
You already get the intersection of the strings or arrays. You can get the union like this:
@res = compare( $str1, $str2 );
$intersection = $res[3];
$union = $res[3].$res[4].$res[5];
@res = compare( \@ary1, \@ary2 );
@intersection = @{$res[3]};
@union = ( @{$res[3]}, @{$res[4]}, @{$res[5]} );
EXPORT
None by default.
AUTHOR
Dr James Freeman <james.freeman@id3.org.uk>
SEE ALSO
perl.