Assuming you want to ignore the 'asdgjhashdnh gsdh sjs klja' stuff and only extract single-digit/alpha-string pairs, this might be a way to go (sorry for the wrap-around):

>perl -wMstrict -le
"my $s = '1 human 2 flower 3 fruits 5 human 6 car 9 flower asdgjhashdn
+h gsdh sjs klja';
 my %hash = $s =~ m{ (\d) \s+ ([[:alpha:]]+) }xmsg;
 use Data::Dumper;
 print Dumper \%hash;
"
$VAR1 = {
          '6' => 'car',
          '1' => 'human',
          '3' => 'fruits',
          '9' => 'flower',
          '2' => 'flower',
          '5' => 'human'
        };
[download]

One problem with
    my ($index, $data)=split (/\s+/,$line);
is that it only captures the first index/data pair, and the example in the OP has several pairs per line. (Update: This is actually pointed out by moritz.) Another problem is that the extraneous data (the 'asdgjhashdnh gsdh sjs klja' stuff) would not be filtered out by a split that simply captured all the field pairs generated.