Re: Re: PLS Advise, Highlighted Text from a Query

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



[...]
what about the percentage how much accurate the result returned??

I'm not sure what kind of keywords or text you're searching, but here's the algorithm i used a while back, for plain-text search:


given:
$query
$text

/*
curve() maps values between 0 and 1 to.. a curve:
  0% ->   0%
 25% ->  43%
 50% ->  75%
 75% ->  93%
100% -> 100%
*/
function curve($x){
  $x = 1 - $x;
  $x = $x * $x;
  return 1 - $x;
}

/*
given a number between 0 and +INFINITY,
strength returns a value between 0 and 1 (0% and 100%)
the higher $n is, the closer to 1 the result.
0 ->   0%       6 -> 85.7%
1 ->  50%       7 -> 87.5%
2 ->  66.6%     8 -> 88.8%
3 ->  75%       9 -> 90%
4 ->  80%      19 -> 95%
5 ->  83.3%   100 -> 99%
*/
function strength($x){
  return ($x / ($x + 1))
}

$count = 0
$keywords = (split $query into array of whole words)
for (each word in $query) {
$n = (number of times $word found in $text);
// add the size of $word to $count, or more depending on $n
$count += (length of $word) * (1 + strength($n));
}
$score = ($count + $count) / (length of $query + length of $text );
$score = curve($score); // repeat a few times if needed, to get decent scores


and there you have it.

-------------------------------------
mikon


-- PHP Database Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php


[Index of Archives]     [PHP Home]     [PHP Users]     [Postgresql Discussion]     [Kernel Newbies]     [Postgresql]     [Yosemite News]

  Powered by Linux