use strict;
use warnings;
use Data::Dumper;

my %hash;
while (<DATA>) {
    my ( $key, $value ) = split;
    $hash{$key} = $value;
} ## end while (<DATA>)
print "Original hash\n", Dumper( \%hash );

my %data_structure;
foreach my $key ( keys %hash ) {
    push @{ $data_structure{ $hash{$key} } }, $key;
} ## end foreach my $key ( keys %hash)
print "Inverted hash\n", Dumper( \%data_structure );

__DATA__
key1, value1
key2, value2
key3, value1
key4, value3
key5, value2
key6, value4
key7, value4
[download]

Original hash
$VAR1 = {
          'key3,' => 'value1',
          'key2,' => 'value2',
          'key4,' => 'value3',
          'key6,' => 'value4',
          'key7,' => 'value4',
          'key5,' => 'value2',
          'key1,' => 'value1'
        };
Inverted hash
$VAR1 = {
          'value4' => [
                        'key6,',
                        'key7,'
                      ],
          'value1' => [
                        'key3,',
                        'key1,'
                      ],
          'value3' => [
                        'key4,'
                      ],
          'value2' => [
                        'key2,',
                        'key5,'
                      ]
        };
[download]

OK. Let's create a second hash whose keys are the array you need.

# Put this after defining %ItemUID.
my %hvalues; # instead of my @hvalues.

...
    if (not exists $ItemUID{$uid}) { $ItemUID{$uid} = $src }

# At end of first while loop, insert this line after the one above.
    if (not exists (%hvalues{src})) { %hvalues{src} = 1; }
[download]

Then keys %hvalues will get you the array you need. You could always say my @hvalues = keys %hvalues; if clarity of that type is needed.

That is the array will need each key that has ch01.03...

If I understand correctly, you have pairs of UID and NAME. There will be several UIDs associated with each name. You want a list of all the unique names and for each name a list of all the UIDs associated with that name. The structure you need for this is a hash of arrays, and you can produce it at the same time you produce your ItemUID hash, as follows:

my %ItemUID = ();
my %ItemName = ();

#Get ItemUID and Item Name (ItemUID is unique to each Test.IData)
{
    my $sql = "SELECT t.IData FROM Tests AS t JOIN Assignments AS a ON
+ a.A
+ssignmentTestID=t.ID JOIN Users AS u ON u.ID=t.UserID WHERE u.Institu
+tionID='004452' AND t.IData REGEXP 'ch01.03.MultipleChoice 049';";

    my $sth = $dbh->prepare($sql);
    $sth->execute();

    while ( my @row = $sth->fetchrow_array() ) {
        my $idata = $row[0];
        while ( $idata =~ /<item([^>]+)/gi ) {
            my $elements = $1;
            my $src;
            my $uid;
            if ( $elements =~ /src=\"\w+\/([^"]+)/i ) { $src = $1 }
            if ( $elements =~ /uid=\"([^"]+)/i ) { $uid = $1 }
            if ( not exists $ItemUID{$uid} ) {
                $ItemUID{$uid} = $src;
                push(@{$ItemName{$src}}, $uid);
            }
        }
    }
    $sth->finish();
}

# you can generate an array of the unique names if you like
my @names = sort keys %ItemNames;
[download]

The hash $ItemName will have keys like 'ch02.03.MultipleChoice 029' and the values will be references to arrays of UIDs.

Hope this helps...

Thanks much the first reply helped and from there I was able to solve the problem. In short the issue was that rather then grouping the results per Section/Instructor it was grouping them by the ItemUID which created over 300 rows returned rather then 18. The issue was partially in my query and the perl code. Thanks again!