예제 #1
0
 /**
  * Static method extractTags
  *
  * @param string  $content  # input content
  * @param int     $top_k    # top_k
  * @param array   $options  # other options
  *
  * @return array $tags
  */
 public static function extractTags($content, $top_k = 20, $options = array())
 {
     $defaults = array('mode' => 'default');
     $options = array_merge($defaults, $options);
     $tags = array();
     $words = Jieba::cut($content);
     $freq = array();
     $total = 0.0;
     foreach ($words as $w) {
         $w = trim($w);
         if (mb_strlen($w, 'UTF-8') < 2) {
             continue;
         }
         if (isset($freq[$w])) {
             $freq[$w] = $freq[$w] + 1.0;
         } else {
             $freq[$w] = 0.0 + 1.0;
         }
         $total = $total + 1.0;
     }
     foreach ($freq as $k => $v) {
         $freq[$k] = $v / $total;
     }
     $tf_idf_list = array();
     foreach ($freq as $k => $v) {
         if (isset(self::$idf_freq[$k])) {
             $idf_freq = self::$idf_freq[$k];
         } else {
             $idf_freq = self::$max_idf;
         }
         $tf_idf_list[$k] = $v * $idf_freq;
     }
     arsort($tf_idf_list);
     $tags = array_slice($tf_idf_list, 0, $top_k, true);
     return $tags;
 }