check_sentence_keyphrases.uni.pl

#!/usr/bin/perl

#Iris, Feb 2017
# script to count the coverage of the typical terms (a list read in from a file) in every sentence
# output: percentage of coverage of the terms in the sentences.

use strict;
use utf8;
use open ":encoding(utf8)";
binmode(STDOUT, ":utf8");
    
#use Text::Match::FastAlternatives;


my $keylist = $ARGV[0]; #file
my $sentencefile = $ARGV[1]; #just a string
my @keywords;
my @bigrams;

#perl Scripts/check_sentence_keyphrases.pl Exp/Train/likelihood_body_harangue.m1l3t3.txt Exp/Dev/Fields/ps_text_harangue.txt 



open(LIST, $keylist) || die "cannot open $keylist\n";
while(<LIST>){

    chomp;
    my ($pattern,$LHOOD,$OCC_0,$FREQ_0,$OCC_1,$FREQ_1) = split /\t+/, $_;

    # only use those patteren with a log0score >20 and 
    # that focus on the strings typical for harangue/peroration
    if(length($pattern)>1 && $LHOOD >20 && $OCC_1>0 )
    {
	if($pattern =~ m/.\s./){ 
	    push @bigrams, $pattern; }
	else{ push @keywords, $pattern;}
    }
}close(LIST);

open(SEN,$sentencefile) || die 'cant open';
while(<SEN>)
{
   my $sentence = $_;
  # print "+ $sentence";
  # foreach my $el (@keywords)
  #  {
  #   if( $sentence =~ /$el/){ print "$el --";}
  # }
   print "\n";
   my $bool=0;
   my $perc=0;
   my @words= split /\s+/, $sentence;
    my $nbrwords =$#words+1;
   LOOP:  foreach my $el (@keywords)
  {
       if( $sentence =~ /$el/){$bool++; }
  }

 # compute the percentage of keywords that match the in sentence.
  if($bool){
	$perc = ( $bool /$nbrwords)*100;
	if($perc >10){
	print "YES $perc = $bool /$nbrwords; - $sentence"; 
   }}
   $bool=0;



}close(SEN);