multi dimensional hash

[Anonymous Monk]

For sure an easy task. But I need some help. I have a file with three columns. I need to organize the information to obtain a data structure that mirrors the following: for each instance of $tag3 my data structure needs to have its frequency (count) and all the unique $tag1 that may come along with $tag3 (same row). ($tag2 is just a control). I have written the following script which creates a multi dimensional hash. The counting is done correctly, but what would be the best way to save all unique values of $tag1? In my script only the last $tag1 seen is kept.

Furthermore I have an ugly "Use of uninitialized value in addition (+) at .\myscript.pl line 14" the first time a $tag3 is inserted in the hash. How can elegantly I prevent it?

Example of result: for my $tag3='conference' the counter should be 3 (correctly done by the script), but I should also register the 2 unique $tag1 which are "conference" and "conferences" (which I don't know the best way to do it).

    #!/usr/bin/perl
    use strict;
    use warnings;
     
    use Data::Dumper qw(Dumper);

    my $line = <DATA>;
    my %hash;
        while($line){
            my ($tag1, $tag2, $tag3) = split(/\t/, $line);
                if ($tag2 =~/NN/) {
                    $hash{$tag3}{frequency} = (($hash{$tag3}{frequency
+})+1);
                    $hash{$tag3}{variants} = $tag1;
                } 
        $line = <DATA>;
        }
    
     
    print Dumper \%hash;

__DATA__
The    DT    the
International    NN    International
for    IN    for
well    NN    well
preparation    NN    preparation
preparation    NN    preparation
in    IN    in
conference    NN    conference
conference    NN    conference
conferences    NN    conference
good    VVG    good