Replace mis-spelt word within a string

Question

I have a basic search script which I'm working on. I want users to be able to enter several keywords. If one of these keywords are mis-spelt, I want to change that word for the search results and/or display a "did you mean ..." message.

I have tried levenshtein but it only seems to work for a single word and doesn't seem very reliable anyway. When using this function, in testing, I came up with this:

<?php
$input = 'ornage ptoato';

$possible_words = explode(' ', trim(strtolower($input)));

foreach($possible_words as $value){

   $words  = array('sony','red', 'indigo','orange','bell','toshiba','potato');

   $shortest = -1;

   foreach ($words as $word) {

       $lev = levenshtein($value, $word);

       if ($lev == 0) {

           $closest = $word;
           $shortest = 0;

           break;
       }

       if ($lev <= $shortest || $shortest < 0) {
           // set the closest match, and shortest distance
           $closest  = $word;
           $shortest = $lev;
       }
   }

}
echo "Input word: $input<br>";
if ($shortest == 0) {
    echo "Exact match found: $closest";
} else {
    echo "Did you mean: $closest?\n";
}

?>

There is foreach within a foreach because I was trying to do it for each word within the search string.

I basically want it to work like Google's "did you mean.." and eBay's "0 results found for one two theer, so we searched for one two three".

What is your question? What are your expected results / obtained results? — emartel, Nov 14 '12 at 23:33

score 1 · Answer 1 · answered Nov 15 '12 at 00:00

Your code needed a little tweaking.

<?php
$input = 'ornage ptoato toshiba butts';
$possible_words = explode(' ', trim(strtolower($input)));
$words = array('sony','red', 'indigo','orange','bell','toshiba','potato');
$threshold = 4;

foreach($possible_words as $value){
    $shortest = -1;
    if( in_array($value, $words) ) {
        printf("Exact match for word: %s\n", $value);
    } else {
        foreach ($words as $word) {
             $lev = levenshtein($value, $word);

             if ($lev <= $shortest || $shortest < 0) {
                  // set the closest match, and shortest distance
                  $closest  = $word;
                  $shortest = $lev;
             }
        }
        if($shortest < $threshold) {
            printf("You typed: %s.\nAssuming you meant: %s\n", $value, $closest);
        } else {
            printf("Could not find acceptable match for: %s\n", $value);
        }
    }
}

The check for acceptable matches needed to go inside the outer loop.
You can use in_array() to search for an exact match before calculating the Levenshtein Distance
You probably only want to match words within reason. [hench $threshold]

Output:

You typed: ornage.
Assuming you meant: orange
You typed: ptoato.
Assuming you meant: potato
Exact match for word: toshiba
Could not find acceptable match for: butts

Replace mis-spelt word within a string

1 Answers1