示例#1
0
 public function setUp()
 {
     $filter = new LowercaseFilter();
     $filter->addFilter(new StopWordFilter(array('the', 'a', 'and', 'is', 'it')));
     $this->tokenizer = new WordTokenizer($filter);
     $this->classifier = new Classifier($this->tokenizer);
     $this->train();
 }
示例#2
0
 public static function createTokenizer()
 {
     $filter = new LowercaseFilter();
     $filter->addFilter(new StopWordFilter(array('the', 'a', 'and', 'is', 'it', 'of', 'to', 'be', 'in')));
     $filter->addFilter(new ShortWordFilter(2));
     $tokenizer = new WordTokenizer($filter);
     return $tokenizer;
 }