in reply to Guessing/Ordering Partial Data

Here's a pretty simple way to score the matching. Convert the array of search terms to a regex alternation and count the number of matches for each datum: 

my @guesses = ('Place De La Gare', 'Rennes');

my $grei = do {  # this case insensitive re added
    local $" = '|';
    qr/@{[map {quotemeta} @guesses]}/i;
};

my $gre = do {
    local $" = '|';
    qr/@{[map {quotemeta} @guesses]}/;
};

my %score;
for (@data) {
    $score{$_} = () = m/($grei)/g;   # edited.
#   $score{$_} += () = m/($gre)/g;   # uncomment to give extra
                                     # credit for exact match
}
# sort keys by value or grep for threshold to pick best matches

[download]

Update: Case insensitivity was being overruled by the compiled $gre. Repaired. Got rid of non-capture grouping and added quotemeta to defang special characters in the data. More: Added the '()=" trick to force array context - that fixes the counts.

After Compline,
Zaxo

Replies are listed 'Best First'.
Re^2: Guessing/Ordering Partial Data
by ropey (Hermit) on Apr 13, 2005 at 04:11 UTC

    I Like Zaxo

    Output after running and using Data::Dumper is

    $VAR1 = {
          'Place de la Gare - Bergerac' => 1,
          'Place De La Gare -  Angers' => 0,
          'Place Thiers -  Nancy' => 0,
          'Place De La Gare -  Rennes' => 1,
          'Place de la Gare' => 1,
          'Place De La Gare - Nevers' => 0,
          'Place De La Gare -  Grenoble' => 0,
          'Place De La Gare 1 -  Grenoble' => 0,
          'Place De La Gare - Angers' => 0,
          'Place de la Gare - Libourne' => 1,
          'Place de la Gare - Moutiers' => 1,
          'Place Mohammed V -  Oujda' => 0,
          'Place De La Gare' => 0,
          'Place Du Chateau - Galerie Marchande Du Rer' => 0,
          'Place de la Gare - Quimper' => 1,
          'Place De La Gare -  Nevers' => 0,
          'Place De La Gare - Rennes' => 1
        };

    [download]

    I would have though that the Rennes match would have a score of 2, it looks to me that its case sensitive (as re-running with the input as 'Place De La Gare' instead of 'Place de la Gare' scores as i would expect. I dont however get why this is seeing as the regex is using -i ?

[reply]
[d/l]

In Section Seekers of Perl Wisdom