text_matching.php

  /**
   * 2つのmecab結果のベクトル内積を計算し、正規化を行う。
   *
   * @param $mecab_rt1 テキスト1
   * @param $mecab_rt2 テキスト2
   */
  public static function textmatching($mecab_rt1, $mecab_rt2){

    // 改行コードで分けて配列に格納する
    $mecab_array1 = explode("\n", $mecab_rt1);
    $mecab_array2 = explode("\n", $mecab_rt2);

    $word_list = array();
    foreach($mecab_array1 as $words1){
      $word_array1 = explode(",", $words1);
      $type = $word_array1[0];
      if($type == 'EOS'){
        break;
      }
      if(count($word_array1) > 5){
        $value = $word_array1[6];
      }else{
        $value = 'no valid word.';
      }

      $type_name = explode("\t", $type);

      if(count($type_name) > 1){
        $prop = $type_name[1];
      }else{
        $prop = $type_name[0];
      }

      if(($prop=='名詞' or $prop=='動詞' or $prop=='形容詞') and $value !='*' ){
        if(array_key_exists($value, $word_list)){
          $word_list[$value] = $word_list[$value] + 1;
        }else{
          $word_list[$value] = '1';
        }
      }
    }

    $word_list2 = array();
    foreach($mecab_array2 as $words2){
      $word_array2 = explode(",", $words2);
      $type = $word_array2[0];
      if($type == 'EOS'){
        break;
      }

      if(count($word_array2) > 5){
        $value = $word_array2[6];
      }else{
        $value = 'no valid word.';
      }


      $type_name = explode("\t", $type);
      if(count($type_name) > 1){
        $prop = $type_name[1];
      }else{
        $prop = $type_name[0];
      }

      if(($prop=='名詞' or $prop=='動詞' or $prop=='形容詞') and $value !='*' ){
        if(array_key_exists($value, $word_list2)){
          $word_list2[$value] = $word_list2[$value] + 1;
        }else{
          $word_list2[$value] = '1';
        }
      }
    }

    if(empty($word_list) or empty($word_list)){
      $error_message = '有効な単語を含みません';
      return $error_message;
    }


    // 各スケールベクトルを作成
    $vector1 = 0;
    foreach($word_list as $word=>$value){
      $vector1 += pow($value, 2);
    }
    $scale1 = sqrt($vector1);

    $vector2 = 0;
    foreach($word_list2 as $word=>$value){
      $vector2 += pow($value, 2);
    }
    $scale2 = sqrt($vector2);

    // 内積をとる
    $inner = 0;
    foreach($word_list as $word=>$value){
      if(array_key_exists($word, $word_list2)){
        $score1 = $value;
        $score2 = $word_list2[$word];

        $inner += ($score1 * $score2);
      }
    }

    $similarity = 0;
    if(($scale1 * $scale2) != 0){
      $similarity = $inner / ($scale1 * $scale2);
    }

    return $similarity;
  }