nb.pl

#!/usr/bin/env perl

use strict;
use utf8;
use Encode;


# 出現確率ファイルを読み込む
# Like.txtの読み込み
open(LIKE, "Likeprob.txt") or die "$!";
open(DISLIKE, "Dislikeprob.txt") or die "$!";

my $like;
my $dislike;
my %like_hash = 0;
my %dislike_hash = 0;


while(<LIKE>){
    chomp;
    my @list = split(/\t/);
    $like_hash{$list[0]} = $list[1];
}
close(LIKE);

while(<DISLIKE>){
    chomp;
    my @list = split(/\t/);
    $dislike_hash{$list[0]} = $list[1];
}
close(DISLIKE);


# LIKE/DISLIKE判定後の配列
my %judge;
my %judge_test;

# 解答データのLike/Dislikeを取得
open(TEST, "train_id.txt") or die "$!";
my @id;
while(<TEST>){
    chomp;
    @_ = split("\t");
    push(@id,@_[0]);
    if(@_[1] =~ /Like/){
	$judge_test{@_[0]} = "Like";
    }elsif(@_[1] =~ /Dislike/){
	$judge_test{@_[0]} = "Dislike";
    }else{
	print("ERROR!\n");
    }
}
close(TEST);

# クローズドテスト用タグの読み込み

# 教師データにはLikeが194個、Dislikeが96個存在する
my $like_num = 194;
my $dislike_num = 96;
my $total = $like_num + $dislike_num;

open(IN,"test.txt") or die "$!";
while(<IN>){
    chomp;
    my $like = log($like_num/$total);
    my $dislike = log($dislike_num/$total);

    my @list = split(/\t/);
    my @taglist = split(",", @list[1]);
    for(my $i=0; $i<$#taglist; $i++){
	if(defined($like_hash{$taglist[$i]})){
	    $like += log($like_hash{$taglist[$i]})
	}else{
	    $like += log(0.000001);
    }
	if(defined($dislike_hash{$taglist[$i]})){
	    $dislike += log($dislike_hash{$taglist[$i]});
	}else{
	    $dislike += log(0.000001);
	}
    }
    # Like/Dislike判定
    if($like > $dislike){
#	print "@list[0] Like\n";
	$judge{@list[0]} = "Like";
    }elsif($dislike > $like){
#	print "$like $dislike Dislike\n";
	$judge{@list[0]} = "Dislike";
    }else{
	print "$like $dislike same\n";
    }
    $like = 0;
    $dislike = 0;
}
close(IN);

my $j=0;
my $k=0;

for(my $i=0; $i<=$#id; $i++){
    if($judge{$id[$i]} eq $judge_test{$id[$i]}){
        $j++;
    }elsif($judge{$id[$i]} ne $judge_test{$id[$i]}){
        $k++;
    }else{
        print("ERROR");
    }
}

print("correct: $j , incorrect: $k \n");
my $p= $j/($j+$k)*100;
print("the precision is $p\%\n");