Extract Keywords From A Text String With PHP

A common issue I have come across in the past is that I have a CMS system, or an old copy of Wordpress, and I need to create a set of keywords to be used in the meta keywords field. To solve this I put together a simple function that runs through a string and picks out the most commonly used words in that list as an array. This is currently set to be 10, but you can change that quite easily.

The first thing the function defines is a list of "stop" words. This is a list of words that occur quite a bit in English text and would therefore interfere with the outcome of the function. The function also uses a variant of the slug function to remove any odd characters that might be in the text.

function extractCommonWords($string){
      $stopWords = array('i','a','about','an','and','are','as','at','be','by','com','de','en','for','from','how','in','is','it','la','of','on','or','that','the','this','to','was','what','when','where','who','will','with','und','the','www');
   
      $string = preg_replace('/\s\s+/i', '', $string); // replace whitespace
      $string = trim($string); // trim the string
      $string = preg_replace('/[^a-zA-Z0-9 -]/', '', $string); // only take alphanumerical characters, but keep the spaces and dashes too…
      $string = strtolower($string); // make it lowercase
   
      preg_match_all('/\b.*?\b/i', $string, $matchWords);
      $matchWords = $matchWords[0];
      
      foreach ( $matchWords as $key=>$item ) {
          if ( $item == '' || in_array(strtolower($item), $stopWords) || strlen($item) <= 3 ) {
              unset($matchWords[$key]);
          }
      }   
      $wordCountArr = array();
      if ( is_array($matchWords) ) {
          foreach ( $matchWords as $key => $val ) {
              $val = strtolower($val);
              if ( isset($wordCountArr[$val]) ) {
                  $wordCountArr[$val]++;
              } else {
                  $wordCountArr[$val] = 1;
              }
          }
      }
      arsort($wordCountArr);
      $wordCountArr = array_slice($wordCountArr, 0, 10);
      return $wordCountArr;
}

The function returns the 10 most commonly occurring words as an array, with the key as the word and the amount of times it occurs as the value. To extract the words just use the implode() function in conjunction with the array_keys() function. To change the number of words returned just alter the value in the third parameter of the array_slice() function near the return statement, currently set to 10. Here is an example of the function in action.

$text = "This is some text. This is some text. Vending Machines are great.";
$words = extractCommonWords($text);
echo implode(',', array_keys($words));

This produces the following output.

some,text,machines,vending

Update

After lots of versions of this code submitted by users I think the most reliable version is this one from Cenk.

function extractKeyWords($string) {
  mb_internal_encoding('UTF-8');
  $stopwords = array();
  $string = preg_replace('/[\pP]/u', '', trim(preg_replace('/\s\s+/iu', '', mb_strtolower($string))));
  $matchWords = array_filter(explode(' ',$string) , function ($item) use ($stopwords) { return !($item == '' || in_array($item, $stopwords) || mb_strlen($item) <= 2 || is_numeric($item));});
  $wordCountArr = array_count_values($matchWords);
  arsort($wordCountArr);
  return array_keys(array_slice($wordCountArr, 0, 10));
}

This will produce the following sort of output.

print implode(',', extractKeyWords("This is some text. This is some text. Vending Machines are great."));
// prints "this,text,some,great,are,vending,machines"

print implode(',', extractKeyWords('হো সয়না সয়না সয়না ওগো সয়না এত জ্বালা সয়না ঘরেতে আমার এ মন রয়না কেন রয়না রয়না'));
// prints "সয়না,রয়না,কেন,আমার,ঘরেতে,ওগো,জ্বালা"

If you use this then you should be aware that it requires PHP 5.5+ due to the use of a closure, but then that shouldn't be a problem :)

Comments

hello can any one help me with a problem? I have a variable containing a html commando like this:

$str="<a href="http://d-grund.dk/" id="home" target="_blank" title="D-Grund.DK"><img alt="D-Grund.DK" border="1" height="90" id="home" src="http://d_grund.dk/images/homelogo.gif" width="166" /></a>";

and i will split it up so i can change the values and sample it again to work with echo $str; sorry for my bad english. i have searched the entire net and nothing is what i need so i hope anyone can help me. my solution is not good because it bet big and time consuming. and i have find some on the internet that look a like that i need but i cant sample it again without destroying anything. sincerely LAT D-Grund.dk Administrator my email is [email protected]

Permalink
I would love to help but I'm not sure what you are trying to do. Passing this string through the extractCommonWords() function probably won't produce any meaningful output as the string is just HTML containing a link and an image.
Name
Philip Norton
Permalink
function extractCommonWords($string){
$stopWords = array('i','a','about','an','and','are','as','at','be','by','com','de','en','for','from','how','in','is','it','la','of','on','or','that','the','this','to','was','what','when','where','who','will','with','und','the','www'); $string = preg_replace('/ss+/i', '', $string); $string = trim($string); // trim the string $string = preg_replace('/[^a-zA-Z0-9 -]/', '', $string); // only take alphanumerical characters, but keep the spaces and dashes too… $string = strtolower($string); // make it lowercase preg_match_all('/\b.*?\b/i', $string, $matchWords); $matchWords = $matchWords[0]; foreach ( $matchWords as $key=&gt;$item ) { if ( $item == '' || in_array(strtolower($item), $stopWords) || strlen($item) &lt;= 3 ) { unset($matchWords[$key]); } } $wordCountArr = array(); if ( is_array($matchWords) ) { foreach ( $matchWords as $key =&gt; $val ) { $val = strtolower($val); if ( isset($wordCountArr[$val]) ) { $wordCountArr[$val]++; } else { $wordCountArr[$val] = 1; } } } arsort($wordCountArr); $wordCountArr = array_slice($wordCountArr, 0, 10); return $wordCountArr; }
}
 
$text = "This is some text. This is some text. Vending Machines are great.";
$words = extractCommonWords($text);

echo implode(',', array_keys($words));

 

Permalink
Thanks for the input, after running the script it was clear that it wouldn't work as there was one too many curly braces in the function. I have removed this in the example so it will run now.
Name
Philip Norton
Permalink
Do you maybe know, what would be the problem, .. when I get my keywords with your script.. I always get letter "p" added right before first keyword and right after last keyword.. os the keyword list looks like this: pkeyword1, keyword2...... keyword10p .. and I dont know what causes this.. thanks
Permalink
Are you passing it a HTML string with p tags at the start and end? Try using strip_tags() first.
Name
Philip Norton
Permalink

Your approach of stripping punctuation was a heck of a lot better than my method. I went the opposite approach and used a stock snippet of code with several preg_replace patterns, except for WordPress curly smart quotes. This is much more elegant and simple. Thanks for that, and here is how I would count word frequency:

$wordCountArray = array_count_values( $matchWords );

No need of a foreach loop. The strtolower() call inside both foreach loops is really not needed either since you have already converted $string to lowercase beforehand.

Permalink
Hey I’ve found your code very useful. Thanks a lot! However I’m finding an issue and that is that I want to divide the number of times a word repeats by the total number of words. The problem I have is that I cannot access each item of the array since the keys change every time I change the text. I was wondering if there is a say around this or if a new dimension can be added to the array so that things like this will work: echo $wordCountArr[2]/$totalwords; Any help will be much appreciated! Thanks in advance! :)
Permalink
By the way, I've also tryed this with now luck... $numrepeats = print_r (array_values($words), true); for ($h=0; $h=5; $h+=1) { $density[$h] = $numrepeats[$h]*2; echo $density[$h]; echo "
"; }
Permalink

@Guillermo I think to get the effect you are looking for you'll need to modify the function in the following way:

function extractCommonWords($string){
      $stopWords = array('i','a','about','an','and','are','as','at','be','by','com','de','en','for','from','how','in','is','it','la','of','on','or','that','the','this','to','was','what','when','where','who','will','with','und','the','www');
   
      $string = preg_replace('/ss+/i', '', $string);
      $string = trim($string); // trim the string
      $string = preg_replace('/[^a-zA-Z0-9 -]/', '', $string); // only take alphanumerical characters, but keep the spaces and dashes too…
      $string = strtolower($string); // make it lowercase
   
      preg_match_all('/\b.*?\b/i', $string, $matchWords);
      $matchWords = $matchWords[0];

      $totalWords = count($matchWords[0]);

      foreach ( $matchWords as $key=&gt;$item ) {
          if ( $item == '' || in_array(strtolower($item), $stopWords) || strlen($item) &lt;= 3 ) {
              unset($matchWords[$key]);
          }
      }
      $wordCountArr = array();
      if ( is_array($matchWords) ) {
          foreach ( $matchWords as $key =&gt; $val ) {
              $val = strtolower($val);
              if ( !isset($wordCountArr[$val]) {
                  $wordCountArr[$val] = array();
              }
              if ( isset($wordCountArr[$val]['count']) ) {
                  $wordCountArr[$val]['count']++;
              } else {
                  $wordCountArr[$val]['count'] = 1;
              }
          }
          arsort($wordCountArr);
          $wordCountArr = array_slice($wordCountArr, 0, 10);
          foreach ( $wordCountArr as $key =&gt; $val) {
              $wordCountArr[$val]['bytotal'] = $wordCountArr[$val]['count'] / $totalWords;
          } 
      }
      return $wordCountArr;
}

This is untested code, but it should get you what you need. :)

Name
Philip Norton
Permalink
Hey thanks for your super-fast response! I get it quite to work (not saying the code is not right, I just didn’t figure out how to implement it). But I’ve found a way around the issue. Thank you very much!
Permalink

Excellent bit of code. I was looking for something to auto-suggest keywords for a content management system. I made the following changes :

1) removed the strtolower call in the foreach loop because the string has already been forced to lower case so all the items will be lower case anyway.

if ( $item == '' || in_array($item, $stopWords) || strlen($item) <= 3 ) {

2) Added a $count parameter to the function so that when it is called you can specify how many words to return.

function extractCommonWords($string,$count){
   ... 

   $wordCountArr = array_slice($wordCountArr, 0, $count);
   return $wordCountArr;
}

The hardest thing to get right is the stop word list.

Permalink

Brilliant. Thanks for this piece of code. I was after doing something very similar so have used this as a base and made a few tweaks here and there so it fits my needs. 

Permalink

Hi

Great code and is working a treat on my site but I have a problem with words containing 'ss'.  The double ss is being removed so a word such as 'password' is being seen as 'paword' in the keywords.

Regards

Permalink

 

Hi, would you please to let me know how I could include spanish characters too?.

For example I need to print: 'tamaño' instead I get: 'tamao'

Thanks in advance
Tolenca

 

Permalink

You need to add those special characters to the regular expression on line 6 of the examples above. You just need to make sure it takes alphanumeric characters as well as the spanish letters. I think that should work.

Name
Philip Norton
Permalink

instead of

$wordCountArr = array();

if ( is_array($matchWords) ) {
  foreach ( $matchWords as $key => $val ) {
    $val = strtolower($val);
    if ( !isset($wordCountArr[$val]) {
      $wordCountArr[$val] = array();
    }
    if ( isset($wordCountArr[$val]['count']) ) {
        $wordCountArr[$val]['count']++;
    } else {
        $wordCountArr[$val]['count'] = 1;
    }
  }
}

I did:

$ignoreOccur = array(1,2);
$wordCountArr = array_diff(array_count_values(explode(" ", matchWords)), $ignoreOccur);

To get all assoc array $wordCoundArr of [word] => [occurences], ignoring words that have occurred 1 or 2 (or specified number of) times.

Permalink

Interesting take on it. I like it! :)

Name
Philip Norton
Permalink

I changed line 13 to not include 'keywords' that are just numbers using the  is_numeric() function

if ( $item == '' || in_array($item, $stopWords) || strlen($item) >= 3 || is_numeric($item) ) {

 

Permalink

Thanks, this is quite useful!

Permalink

Make sure this line:

$string = preg_replace('/ss+/i', '', $string);

reads like this:

$string = preg_replace('/\s\s+/i', '', $string);
Permalink

This script just returns a string of comma separated keywords and supports multibyte characters.
Adjust your stopwords for your locale.

function extractCommonWords($string)
	{	
                $stopwords = array('i','a','about','an','and','are','as','at','be','by','com','de','en','for','from','how','in','is','it','la','of','on','or','that','the','this','to','was','what','when','where','who','will','with','und','the','www');
                $string = mb_strtolower($string); // make it lowercase
		$string = trim(preg_replace('/\s\s+/i', '', $string;)); //remove multiple whitespace
		$string = preg_replace('/[\pP]/', '', $string); // remove punctuation
		$matchWords = array_filter(explode(" ",$string) , function ($item) use ($stopwords) { return !($item == '' || in_array($item, $stopwords) || mb_strlen($item) &lt; 2 || is_numeric($item));});
		$wordCountArr = arsort(array_count_values($matchWords));
		return implode(',', array_keys(array_slice($wordCountArr, 0, 10)));
	}

 

Permalink

Whoops.. Ignore my previous code block.. it was bugged.. try this.

<?php
function extractKeyWords($string)
{
  mb_internal_encoding('UTF-8');
  $stopwords = array('i', 'a', 'about', 'an', 'and', 'are', 'as', 'at', 'be', 'by', 'com', 'de', 'en', 'for', 'from', 'how', 'in', 'is', 'it', 'la', 'of', 'on', 'or', 'that', 'the', 'this', 'to', 'was', 'what', 'when', 'where', 'who', 'will', 'with', 'und', 'the', 'www');
  $string = preg_replace('/[\pP]/', '', trim(preg_replace('/\s\s+/i', '', mb_strtolower(utf8_encode($string)))));
  $matchWords = array_filter(explode(' ', $string), function ($item) use ($stopwords) {
    return !($item == '' || in_array($item, $stopwords) || mb_strlen($item) < 2 || is_numeric($item));
  });
  $wordCountArr = array_count_values($matchWords);
  arsort($wordCountArr);
  return implode(',', array_keys(array_slice($wordCountArr, 0, 10)));
}

 

Permalink
plz help me how to use for arabic text such as "اب ت ث "? thank you
Permalink
I'm afraid I can't answer that as I don't know enough about the Arabic language/alphabet. Sorry :(
Name
Philip Norton
Permalink

Heya. Great script thanks :) Just a note for the guys who want to handle non-anglo characters, if they replace:

$string = preg_replace('/[^a-zA-Z0-9 -]/', '', $string);

with this:

$string = preg_replace('/[^\w\d -]/', '', $string);

It should support extended characters, not sure off the top of my head how much unicode is covered by \w (if any), but it works for a bunch of quick tests I've put through it.

Permalink

to reduce whitespace (followed by another whitespace) within the string, its better to use

preg_replace('/[ ]{2,}/sm', ' ', $text)

 

Permalink
@sammsel - Seems like a good way of doing things, but why is it better? Is there a performance overhead using my method? Also you are only detecting the space character, not all white space character, which might cause problems/
Name
Philip Norton
Permalink
i see if $wordCountArr contain number (ex. year), the output value for it's 0. For return the corect value you must modify $wordCountArr = array_slice($wordCountArr, 0, 10); in $wordCountArr = array_slice($wordCountArr, 0, 10,$preserve_keys=true);
Permalink
Thanks a lot for sharing this tutorial on how to extract keywords from a text string using php. This post really helped me big time. I am so glad that I came to this site. I just hope that these awesome posts will keep on coming.
Permalink
Found this useful, I need a simple help. If there is a word "bmw-x666" It results "bmw,-,x666" but what i want is it shoudnt extract the word containing hyphen. Could anyone help me?
Permalink
How could i set the key phrase of matched words. eg. This is some, Vending Machines are
Permalink
great code Philip - thanks ! is there a way to group the resulting keywords to two or three words ? your eg: some,text,machines,vending to eg: some text, machines vending, some machines, text vending, some vending, text machines... as multiple keywords can give good results in meta working example: samsung, s3, battery, i9300, original, EB900F result: samsung battery, samsung s3 battery, original i9300 battery, EB900F original, battery samsung s3... google accepts up to 10 grouped keywords of one two or three keywords
Permalink
Hi, I try to use above code with a text is 3-In-1. However, why it returns as 3In1? It seems a preg_replace was remove - from a text. Can I skip - removing within a function? Thanks
Permalink

To stop the function removing dashes you need to change the following line

$string = preg_replace('/[^a-zA-Z0-9 -]/', '', $string);

To this:

$string = preg_replace('/[^a-zA-Z0-9 ]/', '', $string);

Line 7 in the code example above.

Name
Philip Norton
Permalink
Output of code is some,text,machines,vending ie only keywords are displayed but if I need frequency count along with keywords for eg. some 3,text 3,machines 2,vending 1. What modifications should I make ?
Permalink
thanks for your source code, it's works well.
Permalink
function extractKeyWords($string)
{
  mb_internal_encoding('UTF-8');
  $stopwords = array();
  $string = preg_replace('/[\pP]/u', '', trim(preg_replace('/\s\s+/iu', '', mb_strtolower($string))));
  $matchWords = array_filter(explode(' ', $string), function ($item) use ($stopwords) {
    return !($item == '' || in_array($item, $stopwords) || mb_strlen($item) <= 2 || is_numeric($item));
  });
  $wordCountArr = array_count_values($matchWords);
  arsort($wordCountArr);
  return implode(',', array_keys(array_slice($wordCountArr, 0, 10)));
}

Now Perfectly For UTF-8

Permalink
Hello Philip, If I want to extract common words from different language (ex. Bengali - হো সয়না সয়না সয়না ওগো সয়না এত জ্বালা সয়না ঘরেতে আমার এ মন রয়না কেন রয়না রয়না) then what changes should I make? Currently with your code it's not working, it gives no output! Please help.
Permalink
Great code Philip - thanks! I just have a little problem, how can i make each keywords a link. Please help
Permalink
It depends on where you want the keywords link to.
Name
Philip Norton
Permalink
Great Code, and works! you can make me for simple life
Permalink
Thank you, the article is very useful for me and I gained knowledge by reading this article
Permalink
Hi, I try to use above code with a text is 3-In-1. However, why it returns as 3In1? information is interesting and a great tutorial.. tnks
Permalink
Thank you for sharing this incredible post with us.
Permalink
great code
Permalink
Very useful guide
Permalink
I want model no of laptop as keyword. plz help. Code is working great! "Dell Inspiron Core i3 6th Gen - (4 GB/1 TB HDD/Linux) 3467 Laptop (14 inch, Black, 1.956 kg)" doesnt extract 3467 from string. Thank you :)
Permalink

This code helped develop a WordPress plugin that generates tag and keywords for each post.

Permalink

Grate post.Thank You for sharing useful information.it is very benefited to the PHP learners.

keep posting more information.

Permalink

Add new comment

The content of this field is kept private and will not be shown publicly.
CAPTCHA
12 + 7 =
Solve this simple math problem and enter the result. E.g. for 1+3, enter 4.
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.