problem with good Code for simular like words

rontgoff

I wrote this code for search between two groups of words but the count is incorrect any ideas

//Check for simular words function

function word_check($word1, $word2){


$synop1 = substr($word1,0,125);  

$word1_array = explode(' ',$synop1);  

$words1 = count($word1_array);

$synop2 = substr($word2,0,125);  

$word2_array = explode(' ',$synop2);  

$words2 = count($word2_array); 

//print "$word1 has <b>$words1</b> words in it.<br>";
//print "$word2 has <b>$words2</b> words in it.<br>";


$word_match_1a = ereg_replace(" ", "|", $word1);
$word_match_1 = ereg_replace("[.]|[?]|[!]|[,]", "", $word_match_1a);

//print "$word_match_1a";

preg_match_all("($word_match_1)", $word2, $out); 

$out_fin = count($out[0]);
//print "<br>$out_fin";
if($out_fin >= '4'){
print "Good Match ($out_fin)";
}

if($out_fin < '4'){
print "No Match ($out_fin)";
}


}

$word1 = "The news for today is Hot, Cold and Fantastic";
$word2 = "The news of today will soon be Hot, Cold and Flantastic";
print "$word1 <br>compared with <br>$word2<br>";
word_check($word1, $word2);
print "<br><br>";

$word1 = "Blah blah go joe";
$word2 = "joe blah go yo yo";

print "$word1 <br>compared with <br>$word2<br>";
word_check($word1, $word2);
print "<br><br>";

$word1 = "I went on a news excursion";
$word2 = "The news I want to tell you";

print "$word1 <br>compared with <br>$word2<br>";
word_check($word1, $word2);

This is a very useful code if it will come out with a correct count.

laserlight

You need to give a positive example, i.e. what is a correct count?

rontgoff

Iit needs to count all words that match.

If you run the code on the first set it returns 6 in reality the count should be 7.

The second set is correct.

The third set gives a count of 3 but should only be 2.

laserlight

oh well, I felt like doing a simple hack before I sleep, so...

/* Compare 2 strings for similiar words
 * Only compare the first 125 characters
 * Words are blocks of word characters separated by whitespace
 * Word characters are defined according to the current locale
 * Comparison is case-sensitive
 * If at least 4 words match, returns true
 * Else returns false
 */
function word_check($str1, $str2) {
	//take the first 125 characters
	$str1 = substr($str1, 0, 125);
	$str2 = substr($str2, 0, 125);

//break $str1 into its constituent words
//first eliminate non-word characters
$str1 = preg_replace("/[^\\w\\t ]/", "", $str1);
//reduce consecutive whitespace into a single space
$str1 = preg_replace("/[\\t ]+/", " ", $str1);
//now split it
$words1 = explode(" ", $str1);

//break $str2 into its constituent words
//first eliminate non-word characters
$str2 = preg_replace("/[^\\w\\t ]/", "", $str2);
//reduce consecutive whitespace into a single space
$str2 = preg_replace("/[\\t ]+/", " ", $str2);
//now split it
$words2 = explode(" ", $str2);

//now we compare arrays
$count = 0;
foreach ($words1 as $word)
	if (in_array($word, $words2))
		$count++;

//return the result
if ($count >= 4)
	return true;
else
	return false;
}

which may or may not do what you want, and may be fairly inefficient depending on what you want.

EDIT:
but only 6 words match in the first example?

rontgoff

Thanks but yeah it is the same problem an incorrect count of matching words. The script works and will be fine for not critical sites. But if you get a solution let me know.

drawmack

could you explain exactly how it is wrong?

rontgoff

it matches to sentences for like words

Example

$sen1 = "the end of the line";
$sen2 = "the line is the end";

Would equal 3 words that match between the two sentences.

They are "the | line | end".

Does that make clear things up.

drawmack

if that's all you want.

WARNING: code is untested and probably needs slight alterations to work.

<?php
function num_same_words_unique($string,$string2,$start,$limit) {

//replace tabs with spaces.
$temp  = eregi_replace("\t",' ',$temp);
$temp2 = eregi_replace("\t",' ',$temp2);

//squash non-word characters - this is probably wrong
$temp  = ereg_replace"!\w",' ',$temp):
$temp2 = ereg_replace"!\w",' ',$temp2):

//trim the modified strings.
$temp  = substr($string,$start,$limit);
$temp2 = substr($string2,$start,$limit);

//create the arrays
$temp_ary1 = explode(' ',$temp);
$temp_ary2 = explode(' ',$temp2);

//create an array that contains one entry for each 
//word that appears in both arrays
$temp_ary3 = array_intersect($temp_ary1,$temp_ary2);

 return count($temp_ary3);
} //end num_same_words_unique

rontgoff

Well Thanks. I will test it and let you know.