http://qs321.pair.com?node_id=1210292


in reply to multi dimensional hash

A hash of hash might be a possible solution. Here I didn't keep a separate frequency but generated a frequency for each $tag1 as the value to the hash.
#!/usr/bin/perl use strict; use warnings; use Data::Dumper qw(Dumper); use List::Util 'sum'; open my $fh, '<', \<<EOF; The DT the International NN International for IN for well NN well preparation NN preparation preparation NN preparation in IN in conference NN conference conference NN conference conferences NN conference good VVG good EOF my %hash; while (<$fh>) { chomp; my ($tag1, $tag2, $tag3) = split /\t/; if ($tag2 =~/NN/) { $hash{$tag3}{$tag1}++; } } print Dumper \%hash; for my $tag (keys %hash) { printf "%s freq: %d\n", $tag, sum values %{ $hash{$tag} }; }
Output:
$VAR1 = { 'well' => { 'well' => 1 }, 'International' => { 'International' => 1 }, 'conference' => { 'conference' => 2, 'conferences' => 1 }, 'preparation' => { 'preparation' => 2 } }; well freq: 1 International freq: 1 conference freq: 3 preparation freq: 2