stopwords.java

一个很不错的词频统计程序,目前只支持英文,中文的本人正在修改中.改好后上传给大家分享

            m_Stopwords.put("needing", dummy);
            m_Stopwords.put("needn't", 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("nigh", dummy);
            m_Stopwords.put("nigher", dummy);
            m_Stopwords.put("nighest", dummy);
            m_Stopwords.put("nisi", dummy);
            m_Stopwords.put("no", dummy);
            m_Stopwords.put("no-one", dummy);
            m_Stopwords.put("nobody", dummy);
            m_Stopwords.put("none", dummy);
            m_Stopwords.put("nor", dummy);
            m_Stopwords.put("not", dummy);
            m_Stopwords.put("nothing", dummy);
            m_Stopwords.put("notwithstanding", dummy);
            m_Stopwords.put("now", dummy);
            m_Stopwords.put("o", dummy);
            m_Stopwords.put("o'er", dummy);
            m_Stopwords.put("of", dummy);
            m_Stopwords.put("off", dummy);
            m_Stopwords.put("often", dummy);
            m_Stopwords.put("on", dummy);
            m_Stopwords.put("once", dummy);
            m_Stopwords.put("one", dummy);
            m_Stopwords.put("oneself", dummy);
            m_Stopwords.put("only", dummy);
            m_Stopwords.put("onto", dummy);
            m_Stopwords.put("open", dummy);
            m_Stopwords.put("or", dummy);
            m_Stopwords.put("other", dummy);
            m_Stopwords.put("otherwise", dummy);
            m_Stopwords.put("ought", dummy);
            m_Stopwords.put("oughtn't", 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("own", dummy);
            m_Stopwords.put("p", dummy);
            m_Stopwords.put("past", dummy);
            m_Stopwords.put("pending", dummy);
            m_Stopwords.put("per", dummy);
            m_Stopwords.put("perhaps", dummy);
            m_Stopwords.put("plus", dummy);
            m_Stopwords.put("possible", dummy);
            m_Stopwords.put("present", dummy);
            m_Stopwords.put("probably", dummy);
            m_Stopwords.put("provided", dummy);
            m_Stopwords.put("providing", dummy);
            m_Stopwords.put("public", dummy);
            m_Stopwords.put("q", dummy);
            m_Stopwords.put("qua", dummy);
            m_Stopwords.put("quite", dummy);
            m_Stopwords.put("r", dummy);
            m_Stopwords.put("rather", dummy);
            m_Stopwords.put("re", dummy);
            m_Stopwords.put("real", dummy);
            m_Stopwords.put("really", dummy);
            m_Stopwords.put("respecting", dummy);
            m_Stopwords.put("right", dummy);
            m_Stopwords.put("round", dummy);
            m_Stopwords.put("s", dummy);
            m_Stopwords.put("same", dummy);
            m_Stopwords.put("sans", dummy);
            m_Stopwords.put("save", dummy);
            m_Stopwords.put("saving", dummy);
            m_Stopwords.put("second", dummy);
            m_Stopwords.put("several", dummy);
            m_Stopwords.put("shall", dummy);
            m_Stopwords.put("shalt", dummy);
            m_Stopwords.put("shan't", dummy);
            m_Stopwords.put("she", dummy);
            m_Stopwords.put("shed", dummy);
            m_Stopwords.put("shell", dummy);
            m_Stopwords.put("she's", dummy);
            m_Stopwords.put("short", dummy);
            m_Stopwords.put("should", dummy);
            m_Stopwords.put("shouldn't", dummy);
            m_Stopwords.put("since", dummy);
            m_Stopwords.put("six", dummy);
            m_Stopwords.put("small", dummy);
            m_Stopwords.put("so", dummy);
            m_Stopwords.put("some", dummy);
            m_Stopwords.put("somebody", dummy);
            m_Stopwords.put("someone", dummy);
            m_Stopwords.put("something", dummy);
            m_Stopwords.put("sometimes", dummy);
            m_Stopwords.put("soon", dummy);
            m_Stopwords.put("special", dummy);
            m_Stopwords.put("still", dummy);
            m_Stopwords.put("such", dummy);
            m_Stopwords.put("summat", dummy);
            m_Stopwords.put("supposing", dummy);
            m_Stopwords.put("sure", dummy);
            m_Stopwords.put("t", dummy);
            m_Stopwords.put("than", dummy);
            m_Stopwords.put("that", dummy);
            m_Stopwords.put("that'd", dummy);
            m_Stopwords.put("that'll", dummy);
            m_Stopwords.put("that's", dummy);
            m_Stopwords.put("the", dummy);
            m_Stopwords.put("thee", dummy);
            m_Stopwords.put("their", dummy);
            m_Stopwords.put("theirs", dummy);
            m_Stopwords.put("their's", dummy);
            m_Stopwords.put("them", dummy);
            m_Stopwords.put("themselves", dummy);
            m_Stopwords.put("then", dummy);
            m_Stopwords.put("there", dummy);
            m_Stopwords.put("there's", dummy);
            m_Stopwords.put("these", dummy);
            m_Stopwords.put("they", dummy);
            m_Stopwords.put("they'd", dummy);
            m_Stopwords.put("they'll", dummy);
            m_Stopwords.put("they're", dummy);
            m_Stopwords.put("they've", dummy);
            m_Stopwords.put("thine", dummy);
            m_Stopwords.put("this", dummy);
            m_Stopwords.put("tho", dummy);
            m_Stopwords.put("those", dummy);
            m_Stopwords.put("thou", dummy);
            m_Stopwords.put("though", dummy);
            m_Stopwords.put("three", dummy);
            m_Stopwords.put("thro'", dummy);
            m_Stopwords.put("through", dummy);
            m_Stopwords.put("throughout", dummy);
            m_Stopwords.put("thru", dummy);
            m_Stopwords.put("thyself", dummy);
            m_Stopwords.put("till", dummy);
            m_Stopwords.put("to", dummy);
            m_Stopwords.put("today", dummy);
            m_Stopwords.put("together", dummy);
            m_Stopwords.put("too", dummy);
            m_Stopwords.put("touching", dummy);
            m_Stopwords.put("toward", dummy);
            m_Stopwords.put("towards", dummy);
            m_Stopwords.put("true", dummy);
            m_Stopwords.put("'twas", dummy);
            m_Stopwords.put("'tween", dummy);
            m_Stopwords.put("'twere", dummy);
            m_Stopwords.put("'twill", dummy);
            m_Stopwords.put("'twixt", dummy);
            m_Stopwords.put("two", dummy);
            m_Stopwords.put("'twould", dummy);
            m_Stopwords.put("u", dummy);
            m_Stopwords.put("under", dummy);
            m_Stopwords.put("underneath", dummy);
            m_Stopwords.put("unless", dummy);
            m_Stopwords.put("unlike", 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("used", dummy);
            m_Stopwords.put("usually", dummy);
            m_Stopwords.put("v", dummy);
            m_Stopwords.put("versus", dummy);
            m_Stopwords.put("very", dummy);
            m_Stopwords.put("via", dummy);
            m_Stopwords.put("vice", dummy);
            m_Stopwords.put("vis-a-vis", dummy);
            m_Stopwords.put("w", dummy);
            m_Stopwords.put("wanna", dummy);
            m_Stopwords.put("wanting", dummy);
            m_Stopwords.put("was", dummy);
            m_Stopwords.put("wasn't", dummy);
            m_Stopwords.put("way", dummy);
            m_Stopwords.put("we", dummy);
            m_Stopwords.put("we'd", dummy);
            m_Stopwords.put("well", dummy);
            m_Stopwords.put("were", dummy);
            m_Stopwords.put("weren't", dummy);
            m_Stopwords.put("wert", dummy);
            m_Stopwords.put("we've", dummy);
            m_Stopwords.put("what", dummy);
            m_Stopwords.put("whatever", dummy);
            m_Stopwords.put("what'll", dummy);
            m_Stopwords.put("what's", dummy);
            m_Stopwords.put("when", dummy);
            m_Stopwords.put("whencesoever", dummy);
            m_Stopwords.put("whenever", dummy);
            m_Stopwords.put("when's", dummy);
            m_Stopwords.put("whereas", dummy);
            m_Stopwords.put("where's", dummy);
            m_Stopwords.put("whether", dummy);
            m_Stopwords.put("which", dummy);
            m_Stopwords.put("whichever", dummy);
            m_Stopwords.put("whichsoever", dummy);
            m_Stopwords.put("while", dummy);
            m_Stopwords.put("whilst", dummy);
            m_Stopwords.put("who", dummy);
            m_Stopwords.put("who'd", dummy);
            m_Stopwords.put("whoever", dummy);
            m_Stopwords.put("whole", dummy);
            m_Stopwords.put("who'll", dummy);
            m_Stopwords.put("whom", dummy);
            m_Stopwords.put("whore", dummy);
            m_Stopwords.put("who's", dummy);
            m_Stopwords.put("whose", dummy);
            m_Stopwords.put("whoso", dummy);
            m_Stopwords.put("whosoever", dummy);
            m_Stopwords.put("will", dummy);
            m_Stopwords.put("with", dummy);
            m_Stopwords.put("within", dummy);
            m_Stopwords.put("without", dummy);
            m_Stopwords.put("wont", dummy);
            m_Stopwords.put("would", dummy);
            m_Stopwords.put("wouldn't", dummy);
            m_Stopwords.put("wouldst", dummy);
            m_Stopwords.put("x", dummy);
            m_Stopwords.put("y", dummy);
            m_Stopwords.put("ye", dummy);
            m_Stopwords.put("yet", dummy);
            m_Stopwords.put("you", dummy);
            m_Stopwords.put("you'd", dummy);
            m_Stopwords.put("you'll", dummy);
            m_Stopwords.put("your", dummy);
            m_Stopwords.put("you're", dummy);
            m_Stopwords.put("yours", dummy);
            m_Stopwords.put("yourself", dummy);
            m_Stopwords.put("yourselves", dummy);
            m_Stopwords.put("you've", dummy);
            m_Stopwords.put("z", dummy);
        }
    }

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

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