/**
* Get the term frequency
* @param DocumentAbstract $document - the document to evaluate
* @param string $token The token to look for
* @param int $mode The type of term frequency to use
* @return int|float
*/
public function getTermFrequency(DocumentAbstract $document, $token, $mode = 1)
{
$freqDist = new FreqDist($document->getDocumentData());
$keyValuesByWeight = $freqDist->getKeyValuesByFrequency();
//The token does not exist in the document
if (!isset($keyValuesByWeight[$token])) {
return 0;
}
switch ($mode) {
case self::BOOLEAN_MODE:
//a test was already performed if the token exists in the document
//just return true
return 1;
case self::LOGARITHMIC_MODE:
return log($keyValuesByWeight[$token] + 1);
case self::AUGMENTED_MODE:
//FreqDist getKeyValuesByFrequency is already sorted
//in ascending order
$maxFrequency = current($keyValuesByWeight);
return 0.5 + 0.5 * $keyValuesByWeight[$token] / $maxFrequency;
return $keyValuesByWeight;
case self::FREQUENCY_MODE:
default:
return $keyValuesByWeight[$token];
}
}
/**
*
* @param array $tokens
* @param mixed $id
* @param DateTime $createdOn
* @param array $metadata
*/
public function __construct(array $tokens, $id = null, DateTime $createdOn = null, $metadata = [])
{
parent::__construct($tokens, null);
$this->id = $id ?: ++self::$counter;
$this->createdOn = $createdOn ?: new DateTime();
$this->metadata = $metadata;
}