Using Perl to sort an array using values that occur after a RegEx or symbol

I have an array:

@all (
<side.effect signif="life.threat">myocardial infarction</side.effect>
<side.effect signif="life.threat">hypersensitivity reactions</side.effect>
<side.effect signif="life.threat">lactic acidosis</side.effect>
<side.effect signif="most.freq">vomiting</side.effect>
<side.effect signif="most.freq">diarrhea</side.effect>
);

I want to sort the array on the values after the opening XML tags/attributes (">) to produce this output:

<side.effect signif="most.freq">diarrhea</side.effect>
<side.effect signif="life.threat">hypersensitivity reactions</side.effect>
<side.effect signif="life.threat">lactic acidosis</side.effect>
<side.effect signif="life.threat">myocardial infarction</side.effect>
<side.effect signif="most.freq">vomiting</side.effect>

I cannot convert it to a hash as that would eliminate the tags due to replication.
I tried this but it doesn't sort them:

    my @sorted_all = sort {
    my ($aa, $bb) = map { (split)[1] } $a, $b;
    $bb <=> $aa;
} @all;

Using Sort::Key

use strict;
use warnings;
use feature qw(say);

use Sort::Key qw(keysort);

my @all = ( 
    q{<side.effect signif="life.threat">myocardial infarctio</side.effect>},
    q{<side.effect signif="life.threat">hypersensitivity reations</side.effect>},
    q{<side.effect signif="life.threat">lactic acidosis</sid.effect>},
    q{<side.effect signif="most.freq">vomiting</side.effect>},
    q{<side.effect signif="most.freq">diarrhea</side.effect>},
);


my @sorted = keysort { ( /">(.+?)<\// )[0] } @all;  

say for @sorted;

The library uses Schwartzian Transform when needed, to first build the comparison patterns for all items (and not re-do it at each pair comparison). I copied input as given, typos and all.

Using a regex to parse an XML tag relies on this very specific input format. If there are going to be variations in format then please use a proper XML parser, like XML::LibXML. For example

use XML::LibXML;

my $parser = XML::LibXML->new;

my @sorted = keysort {
    $parser -> parse_string($_)
        -> findnodes('side.effect') -> [0]
        -> textContent
} @all;

For this code see XML::LibXML::Parser and XML::LibXML::Node. This library comes with a lot more documentation, see the top document linked at first mention.

For this to work I had to correct the typo sid.effect in one node so to have a valid XML.