stopwords.java

Java 编写的多种数据挖掘算法 包括聚类、分类、预处理等

      m_Stopwords.put("many", dummy);
      m_Stopwords.put("may", dummy);
      m_Stopwords.put("maybe", dummy);
      m_Stopwords.put("me", dummy);
      m_Stopwords.put("mean", dummy);
      m_Stopwords.put("meanwhile", dummy);
      m_Stopwords.put("merely", dummy);
      m_Stopwords.put("might", dummy);
      m_Stopwords.put("more", dummy);
      m_Stopwords.put("moreover", dummy);
      m_Stopwords.put("most", dummy);
      m_Stopwords.put("mostly", dummy);
      m_Stopwords.put("much", dummy);
      m_Stopwords.put("must", dummy);
      m_Stopwords.put("my", dummy);
      m_Stopwords.put("myself", dummy);
      m_Stopwords.put("n", dummy);
      m_Stopwords.put("name", dummy);
      m_Stopwords.put("namely", dummy);
      m_Stopwords.put("nd", dummy);
      m_Stopwords.put("near", dummy);
      m_Stopwords.put("nearly", dummy);
      m_Stopwords.put("necessary", dummy);
      m_Stopwords.put("need", dummy);
      m_Stopwords.put("needs", dummy);
      m_Stopwords.put("neither", dummy);
      m_Stopwords.put("never", dummy);
      m_Stopwords.put("nevertheless", dummy);
      m_Stopwords.put("new", dummy);
      m_Stopwords.put("next", dummy);
      m_Stopwords.put("nine", dummy);
      m_Stopwords.put("no", dummy);
      m_Stopwords.put("nobody", dummy);
      m_Stopwords.put("non", dummy);
      m_Stopwords.put("none", dummy);
      m_Stopwords.put("noone", dummy);
      m_Stopwords.put("nor", dummy);
      m_Stopwords.put("normally", dummy);
      m_Stopwords.put("not", dummy);
      m_Stopwords.put("nothing", dummy);
      m_Stopwords.put("novel", dummy);
      m_Stopwords.put("now", dummy);
      m_Stopwords.put("nowhere", dummy);
      m_Stopwords.put("o", dummy);
      m_Stopwords.put("obviously", dummy);
      m_Stopwords.put("of", dummy);
      m_Stopwords.put("off", dummy);
      m_Stopwords.put("often", dummy);
      m_Stopwords.put("oh", dummy);
      m_Stopwords.put("ok", dummy);
      m_Stopwords.put("okay", dummy);
      m_Stopwords.put("old", dummy);
      m_Stopwords.put("on", dummy);
      m_Stopwords.put("once", dummy);
      m_Stopwords.put("one", dummy);
      m_Stopwords.put("ones", dummy);
      m_Stopwords.put("only", dummy);
      m_Stopwords.put("onto", dummy);
      m_Stopwords.put("or", dummy);
      m_Stopwords.put("other", dummy);
      m_Stopwords.put("others", dummy);
      m_Stopwords.put("otherwise", dummy);
      m_Stopwords.put("ought", dummy);
      m_Stopwords.put("our", dummy);
      m_Stopwords.put("ours", dummy);
      m_Stopwords.put("ourselves", dummy);
      m_Stopwords.put("out", dummy);
      m_Stopwords.put("outside", dummy);
      m_Stopwords.put("over", dummy);
      m_Stopwords.put("overall", dummy);
      m_Stopwords.put("own", dummy);
      m_Stopwords.put("p", dummy);
      m_Stopwords.put("particular", dummy);
      m_Stopwords.put("particularly", dummy);
      m_Stopwords.put("per", dummy);
      m_Stopwords.put("perhaps", dummy);
      m_Stopwords.put("placed", dummy);
      m_Stopwords.put("please", dummy);
      m_Stopwords.put("plus", dummy);
      m_Stopwords.put("possible", dummy);
      m_Stopwords.put("presumably", dummy);
      m_Stopwords.put("probably", dummy);
      m_Stopwords.put("provides", dummy);
      m_Stopwords.put("q", dummy);
      m_Stopwords.put("que", dummy);
      m_Stopwords.put("quite", dummy);
      m_Stopwords.put("qv", dummy);
      m_Stopwords.put("r", dummy);
      m_Stopwords.put("rather", dummy);
      m_Stopwords.put("rd", dummy);
      m_Stopwords.put("re", dummy);
      m_Stopwords.put("really", dummy);
      m_Stopwords.put("reasonably", dummy);
      m_Stopwords.put("regarding", dummy);
      m_Stopwords.put("regardless", dummy);
      m_Stopwords.put("regards", dummy);
      m_Stopwords.put("relatively", dummy);
      m_Stopwords.put("respectively", dummy);
      m_Stopwords.put("right", dummy);
      m_Stopwords.put("s", dummy);
      m_Stopwords.put("said", dummy);
      m_Stopwords.put("same", dummy);
      m_Stopwords.put("saw", dummy);
      m_Stopwords.put("say", dummy);
      m_Stopwords.put("saying", dummy);
      m_Stopwords.put("says", dummy);
      m_Stopwords.put("second", dummy);
      m_Stopwords.put("secondly", dummy);
      m_Stopwords.put("see", dummy);
      m_Stopwords.put("seeing", dummy);
      m_Stopwords.put("seem", dummy);
      m_Stopwords.put("seemed", dummy);
      m_Stopwords.put("seeming", dummy);
      m_Stopwords.put("seems", dummy);
      m_Stopwords.put("seen", dummy);
      m_Stopwords.put("self", dummy);
      m_Stopwords.put("selves", dummy);
      m_Stopwords.put("sensible", dummy);
      m_Stopwords.put("sent", dummy);
      m_Stopwords.put("serious", dummy);
      m_Stopwords.put("seriously", dummy);
      m_Stopwords.put("seven", dummy);
      m_Stopwords.put("several", dummy);
      m_Stopwords.put("shall", dummy);
      m_Stopwords.put("she", dummy);
      m_Stopwords.put("should", dummy);
      m_Stopwords.put("since", dummy);
      m_Stopwords.put("six", dummy);
      m_Stopwords.put("so", dummy);
      m_Stopwords.put("some", dummy);
      m_Stopwords.put("somebody", dummy);
      m_Stopwords.put("somehow", dummy);
      m_Stopwords.put("someone", dummy);
      m_Stopwords.put("something", dummy);
      m_Stopwords.put("sometime", dummy);
      m_Stopwords.put("sometimes", dummy);
      m_Stopwords.put("somewhat", dummy);
      m_Stopwords.put("somewhere", dummy);
      m_Stopwords.put("soon", dummy);
      m_Stopwords.put("sorry", dummy);
      m_Stopwords.put("specified", dummy);
      m_Stopwords.put("specify", dummy);
      m_Stopwords.put("specifying", dummy);
      m_Stopwords.put("still", dummy);
      m_Stopwords.put("sub", dummy);
      m_Stopwords.put("such", dummy);
      m_Stopwords.put("sup", dummy);
      m_Stopwords.put("sure", dummy);
      m_Stopwords.put("t", dummy);
      m_Stopwords.put("take", dummy);
      m_Stopwords.put("taken", dummy);
      m_Stopwords.put("tell", dummy);
      m_Stopwords.put("tends", dummy);
      m_Stopwords.put("th", dummy);
      m_Stopwords.put("than", dummy);
      m_Stopwords.put("thank", dummy);
      m_Stopwords.put("thanks", dummy);
      m_Stopwords.put("thanx", dummy);
      m_Stopwords.put("that", dummy);
      m_Stopwords.put("thats", dummy);
      m_Stopwords.put("the", dummy);
      m_Stopwords.put("their", dummy);
      m_Stopwords.put("theirs", dummy);
      m_Stopwords.put("them", dummy);
      m_Stopwords.put("themselves", dummy);
      m_Stopwords.put("then", dummy);
      m_Stopwords.put("thence", dummy);
      m_Stopwords.put("there", dummy);
      m_Stopwords.put("thereafter", dummy);
      m_Stopwords.put("thereby", dummy);
      m_Stopwords.put("therefore", dummy);
      m_Stopwords.put("therein", dummy);
      m_Stopwords.put("theres", dummy);
      m_Stopwords.put("thereupon", dummy);
      m_Stopwords.put("these", dummy);
      m_Stopwords.put("they", dummy);
      m_Stopwords.put("think", dummy);
      m_Stopwords.put("third", dummy);
      m_Stopwords.put("this", dummy);
      m_Stopwords.put("thorough", dummy);
      m_Stopwords.put("thoroughly", dummy);
      m_Stopwords.put("those", dummy);
      m_Stopwords.put("though", dummy);
      m_Stopwords.put("three", dummy);
      m_Stopwords.put("through", dummy);
      m_Stopwords.put("throughout", dummy);
      m_Stopwords.put("thru", dummy);
      m_Stopwords.put("thus", dummy);
      m_Stopwords.put("to", dummy);
      m_Stopwords.put("together", dummy);
      m_Stopwords.put("too", dummy);
      m_Stopwords.put("took", dummy);
      m_Stopwords.put("toward", dummy);
      m_Stopwords.put("towards", dummy);
      m_Stopwords.put("tried", dummy);
      m_Stopwords.put("tries", dummy);
      m_Stopwords.put("truly", dummy);
      m_Stopwords.put("try", dummy);
      m_Stopwords.put("trying", dummy);
      m_Stopwords.put("twice", dummy);
      m_Stopwords.put("two", dummy);
      m_Stopwords.put("u", dummy);
      m_Stopwords.put("un", dummy);
      m_Stopwords.put("under", dummy);
      m_Stopwords.put("unfortunately", dummy);
      m_Stopwords.put("unless", dummy);
      m_Stopwords.put("unlikely", dummy);
      m_Stopwords.put("until", dummy);
      m_Stopwords.put("unto", dummy);
      m_Stopwords.put("up", dummy);
      m_Stopwords.put("upon", dummy);
      m_Stopwords.put("us", dummy);
      m_Stopwords.put("use", dummy);
      m_Stopwords.put("used", dummy);
      m_Stopwords.put("useful", dummy);
      m_Stopwords.put("uses", dummy);
      m_Stopwords.put("using", dummy);
      m_Stopwords.put("usually", dummy);
      m_Stopwords.put("uucp", dummy);
      m_Stopwords.put("v", dummy);
      m_Stopwords.put("value", dummy);
      m_Stopwords.put("various", dummy);
      m_Stopwords.put("ve", dummy); //added to avoid words like I've,you've etc.
      m_Stopwords.put("very", dummy);
      m_Stopwords.put("via", dummy);
      m_Stopwords.put("viz", dummy);
      m_Stopwords.put("vs", dummy);
      m_Stopwords.put("w", dummy);
      m_Stopwords.put("want", dummy);
      m_Stopwords.put("wants", dummy);
      m_Stopwords.put("was", dummy);
      m_Stopwords.put("way", dummy);
      m_Stopwords.put("we", dummy);
      m_Stopwords.put("welcome", dummy);
      m_Stopwords.put("well", dummy);
      m_Stopwords.put("went", dummy);
      m_Stopwords.put("were", dummy);
      m_Stopwords.put("what", dummy);
      m_Stopwords.put("whatever", dummy);
      m_Stopwords.put("when", dummy);
      m_Stopwords.put("whence", dummy);
      m_Stopwords.put("whenever", dummy);
      m_Stopwords.put("where", dummy);
      m_Stopwords.put("whereafter", dummy);
      m_Stopwords.put("whereas", dummy);
      m_Stopwords.put("whereby", dummy);
      m_Stopwords.put("wherein", dummy);
      m_Stopwords.put("whereupon", dummy);
      m_Stopwords.put("wherever", dummy);
      m_Stopwords.put("whether", dummy);
      m_Stopwords.put("which", dummy);
      m_Stopwords.put("while", dummy);
      m_Stopwords.put("whither", dummy);
      m_Stopwords.put("who", dummy);
      m_Stopwords.put("whoever", dummy);
      m_Stopwords.put("whole", dummy);
      m_Stopwords.put("whom", dummy);
      m_Stopwords.put("whose", dummy);
      m_Stopwords.put("why", dummy);
      m_Stopwords.put("will", dummy);
      m_Stopwords.put("willing", dummy);
      m_Stopwords.put("wish", dummy);
      m_Stopwords.put("with", dummy);
      m_Stopwords.put("within", dummy);
      m_Stopwords.put("without", dummy);
      m_Stopwords.put("wonder", dummy);
      m_Stopwords.put("would", dummy);
      m_Stopwords.put("would", dummy);
      m_Stopwords.put("x", dummy);
      m_Stopwords.put("y", dummy);
      m_Stopwords.put("yes", dummy);
      m_Stopwords.put("yet", dummy);
      m_Stopwords.put("you", dummy);
      m_Stopwords.put("your", dummy);
      m_Stopwords.put("yours", dummy);
      m_Stopwords.put("yourself", dummy);
      m_Stopwords.put("yourselves", dummy);
      m_Stopwords.put("z", dummy);
      m_Stopwords.put("zero", dummy);

    }
  }
  
  /** 
   * Returns true if the given string is a stop word.
   */
  public static boolean isStopword(String str) {

    return m_Stopwords.containsKey(str.toLowerCase());
  }
}