public function __invoke($data)
{
foreach ($data as $category => $documents) {
foreach ($documents as $index => $document) {
if ($this->documentNormalizer) {
$document = $this->documentNormalizer->normalize($document);
}
$tokens = $this->tokenizer->tokenize($document);
if ($this->tokenNormalizer) {
$tokens = $this->tokenNormalizer->normalize($tokens);
}
$data[$category][$index] = $tokens;
}
}
return $data;
}
/**
* @inheritdoc
*/
public function classify($document)
{
$results = array();
if ($this->documentNormalizer) {
$document = $this->documentNormalizer->normalize($document);
}
$tokens = $this->tokenizer->tokenize($document);
if ($this->tokenNormalizer) {
$tokens = $this->tokenNormalizer->normalize($tokens);
}
$tokens = array_count_values($tokens);
$weights = $this->preparedModel()->getModel();
foreach (array_keys($weights) as $category) {
$results[$category] = 0;
foreach ($tokens as $token => $count) {
if (array_key_exists($token, $weights[$category])) {
$results[$category] += $count * $weights[$category][$token];
}
}
}
asort($results, SORT_NUMERIC);
$category = key($results);
$value = array_shift($results);
if ($value === array_shift($results)) {
return false;
} else {
return $category;
}
}
/**
* Formats the document for use in \SVMModel
* @param string $document
* @param \Camspiers\StatisticalClassifier\Model\SVMModel $model
* @return array
*/
protected function prepareDocument($document, SVMModel $model)
{
$tokenMap = $model->getTokenMap();
$data = array();
if ($this->documentNormalizer) {
$document = $this->documentNormalizer->normalize($document);
}
$tokens = $this->tokenizer->tokenize($document);
if ($this->tokenNormalizer) {
$tokens = $this->tokenNormalizer->normalize($tokens);
}
$tokenCounts = array_count_values($tokens);
foreach ($tokenCounts as $token => $value) {
if (isset($tokenMap[$token])) {
$data[$tokenMap[$token]] = $value;
}
}
ksort($data, SORT_NUMERIC);
return $data;
}
