File Coverage

blib/lib/Lingua/EN/WSD/CorpusBased.pm

Criterion	Covered	Total	%
statement	10	12	83.3
branch			n/a
condition			n/a
subroutine	4	4	100.0
pod			n/a
total	14	16	87.5

package Lingua::EN::WSD::CorpusBased;

our $VERSION = "0.11";

use warnings;
use strict;
use String::Util qw(crunch);
use stem;
use Lingua::EN::WSD::CorpusBased::Corpus;

sub new {
    my $class = shift;

    my %args = ('debug' => 0,
		'strict' => 0,
		'stem' => 1,
		'hypernyms' => 1,
		'hyponyms' => 1,
		@_);
    return -1 if (ref($args{'wnref'}) ne 'WordNet::QueryData');
    return -1 if (ref($args{'cref'}) ne 'Lingua::EN::WSD::CorpusBased::Corpus');

    my $this = { 'wn' => $args{'wnref'},
		 'corpus' => $args{'cref'},
		 'debug' => $args{'debug'},
		 'strict' => $args{'strict'},
		 'flexibility' => $args{'flexibility'},
		 'stem' => $args{'stem'},
		 'stemmer' => stem->new($args{'wnref'}),
		 'hype' => $args{'hypernyms'},
		 'hypo' => $args{'hyponyms'}};
    return bless $this, $class;
}

sub init {
    my $self = shift;
    my %args = @_;
    my $term = $args{'term'};
    my @t = split / +/, lc($term);
    $self->{'term'} = \@t;
    $self->{'size'} = scalar @t;
    $self->{'inWN'} = $self->{'wn'}->queryWord($self->term,"syns");
    $self->{'wnTerm'} = crunch($self->{'stemmer'}->stemString(($self->{'inWN'}?$self->term:$self->head)));
    if ($self->{'debug'} > 0) {
	print STDERR 'CorpusBased object initialized (term="'.join(' ',@{ $self->{'term'} }).'", ';
	print STDERR 'size="'.$self->{'size'}.'", ';
	print STDERR 'inWN="'.$self->{'inWN'}.'", ';
	print STDERR 'wnTerm="'.$self->{'wnTerm'}.")\n";
    }
}

sub ready {
    my $self = shift;
    return 0 if (! exists($self->{'term'}));
    return 0 if ($self->{'term'} eq '');
    return 1;
}

sub term {
    my $self = shift;
    return -1 if (! $self->ready);
    return lc(join("_",@{$self->{'term'}}));
}

sub head {
    my $self = shift;
    return -1 if (! $self->ready);
    return lc($self->{'term'}->[$self->{'size'} - 1]);
}

sub term_replace {
    my $self = shift;
    my $replacement = shift;
    return -1 if (! $self->ready);
    my @t = @{$self->{'term'}};
    pop(@t);
    push(@t,split(/_/,$replacement));
    return \@t;
}


sub synsets {
    my $self = shift;
    my %returnlist = ();
    return [] if (! $self->ready);
    my $query = $self->{'wnTerm'};
    my @senses = $self->{'wn'}->queryWord($query,"syns");
    foreach my $word_pos (@senses) {
	foreach my $word_pos_num ($self->{'wn'}->querySense($word_pos,"syns")) {
	    $returnlist{$word_pos_num} = 1;
	};
    };
    my @t = keys %returnlist;

    return \@t;
}

sub synonyms {
    my ($self,$synset) = @_;
    return () if ((! defined $synset) or $synset eq '');
    return (map
	    { /([^\#]*)\#[nva]\#\d+/; $1 }
	    $self->{'wn'}->querySense($synset,"syns"));
}

sub hypernyms {
    my ($self,$synset) = @_;
    return () if ((! defined $synset) or $synset eq '');
    return $self->{'wn'}->querySense($synset,"hype");
}

sub hyponyms {
    my ($self,$synset) = @_;
    return () if ((! defined $synset) or $synset eq '');
    return $self->{'wn'}->querySense($synset,"hypo");
}

sub count {
    my $self = shift;
    return -1 if ((! defined $self->{'corpus'}) or
		  (! exists $self->{'corpus'}));
    return $self->{'corpus'}->count(@_);
}

sub v {
    my ($self,$synset) = @_;
    return -1 if (! defined $synset or
		  $synset eq '');
    my $sum = 0;
    foreach my $synonym ($self->synonyms($synset)) {
	if ($self->{'inWN'}) {
	    $sum += $self->count(split(/_/,$synonym));
	} else {
	    $sum += $self->count(@{$self->term_replace($synonym)});
	}
    };

    # hypernyms
    if ($self->{'hype'}) {
	foreach my $hypernym ($self->hypernyms($synset)) {
	    foreach my $synonym ($self->synonyms($hypernym)) {
		if ($self->{'inWN'}) {
		    $sum += $self->count(split(/_/,$synonym));
		} else {
		    $sum += $self->count(@{$self->term_replace($synonym)});
		}
	    };
	};
    };

    # hyponyms
    if ($self->{'hypo'}) {
	foreach my $hyponym ($self->hyponyms($synset)) {
	    foreach my $synonym ($self->synonyms($hyponym)) {
		if ($self->{'inWN'}) {
		    $sum += $self->count(split(/_/,$synonym));
		} else {
		    $sum += $self->count(@{$self->term_replace($synonym)});
		}
	    };
	};
    };
    return $sum;
}

sub sense {
    my $self = shift;
    my $max = 0;
    return [] if ($self->{'wnTerm'} eq '');
    my @maxsynsets = ();
    foreach my $synset (@{$self->synsets}) {
	print STDERR '  '.$synset."\n" if ($self->{'debug'} > 0);
	my $value = $self->v($synset)."\n";
	if ($value > $max) {
	    @maxsynsets = ($synset);
	    $max = $value;
	} elsif ($value == $max) {
	    push(@maxsynsets, $synset);
	};
    }
    if (scalar @maxsynsets == scalar @{ $self->synsets } and
	$self->{'strict'}) {
	return [];
    }
    return \@maxsynsets;
}

sub wsd {
    my ($self,$term) = @_;
    return -1 if ($term eq '');
    print STDERR "Doing WSD for term '$term'\n" if ($self->{'debug'} > 0);
    $self->init('term' => $term);
    return $self->sense;
}

sub debug {
    my $self = shift;
    return $self->{'debug'};
}

1;

__END__