One way of approaching a problem of this kind is to find an effective way to represent the data. First, I'll assume your files are small enough to hold in memory. Putting files into an array of arrays:

my $ifs = '|'; # or whatever

open FH, "<$file1path" or die $1;
my @file1data = map {[split $ifs, $_]} <FH>;
close(FH( or die $!;

open FH, "<$file2path" or die $1;
my @file2data = map {[split $ifs, $_]} <FH>;
close(FH) or die $!;
[download]

Now we introduce a third array of arrays, for scores. It will have a row for each record in file1, and a column for each in file2. The number of matches will be stored in each:

my $lastrec = scalar( @{$file1data[0]}) - 1;

my @scores = map {
    my $d1 = $_;         # arrayref of a record from file11
    [ map {              # an arrayref with score for each record from
+ file2
        my $d2 = $_;     # arrayref of a record from file12
        scalar grep {    # count stringy matches 
            $d1->[$_] eq $d2->[$_];
        } 0..$lastrec;
    } @file2data ];
} @file1data;
[download]

At this point you can scan a row of @scores for the indexes of best match, or whatever statistics you want. This could have been done as well in nested for loops, but map is fun to play with. Warning, it compiles, but it's untested code.

After Compline,
Zaxo