lovinsstemmer.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

package edu.udo.cs.wvtool.external;

import java.io.Serializable;
import java.util.HashMap;

import edu.udo.cs.wvtool.util.WVToolLogger;

/**
 * Implements the Lovins stemmer.
 * 
 * @author Eibe Frank ([e-mail])
 * @version 1.0
 * 
 */
public class LovinsStemmer implements Serializable {

    /**
     * C version compatibility mode (emulates bugs in original C implementation)
     */
    private static boolean m_CompMode = true;

    /** The hash tables containing the list of endings. */
    private static HashMap m_l11 = null;

    private static HashMap m_l10 = null;

    private static HashMap m_l9 = null;

    private static HashMap m_l8 = null;

    private static HashMap m_l7 = null;

    private static HashMap m_l6 = null;

    private static HashMap m_l5 = null;

    private static HashMap m_l4 = null;

    private static HashMap m_l3 = null;

    private static HashMap m_l2 = null;

    private static HashMap m_l1 = null;

    static {

        m_l11 = new HashMap();
        m_l11.put("alistically", "B");
        m_l11.put("arizability", "A");
        m_l11.put("izationally", "B");
        m_l10 = new HashMap();
        m_l10.put("antialness", "A");
        m_l10.put("arisations", "A");
        m_l10.put("arizations", "A");
        m_l10.put("entialness", "A");
        m_l9 = new HashMap();
        m_l9.put("allically", "C");
        m_l9.put("antaneous", "A");
        m_l9.put("antiality", "A");
        m_l9.put("arisation", "A");
        m_l9.put("arization", "A");
        m_l9.put("ationally", "B");
        m_l9.put("ativeness", "A");
        m_l9.put("eableness", "E");
        m_l9.put("entations", "A");
        m_l9.put("entiality", "A");
        m_l9.put("entialize", "A");
        m_l9.put("entiation", "A");
        m_l9.put("ionalness", "A");
        m_l9.put("istically", "A");
        m_l9.put("itousness", "A");
        m_l9.put("izability", "A");
        m_l9.put("izational", "A");
        m_l8 = new HashMap();
        m_l8.put("ableness", "A");
        m_l8.put("arizable", "A");
        m_l8.put("entation", "A");
        m_l8.put("entially", "A");
        m_l8.put("eousness", "A");
        m_l8.put("ibleness", "A");
        m_l8.put("icalness", "A");
        m_l8.put("ionalism", "A");
        m_l8.put("ionality", "A");
        m_l8.put("ionalize", "A");
        m_l8.put("iousness", "A");
        m_l8.put("izations", "A");
        m_l8.put("lessness", "A");
        m_l7 = new HashMap();
        m_l7.put("ability", "A");
        m_l7.put("aically", "A");
        m_l7.put("alistic", "B");
        m_l7.put("alities", "A");
        m_l7.put("ariness", "E");
        m_l7.put("aristic", "A");
        m_l7.put("arizing", "A");
        m_l7.put("ateness", "A");
        m_l7.put("atingly", "A");
        m_l7.put("ational", "B");
        m_l7.put("atively", "A");
        m_l7.put("ativism", "A");
        m_l7.put("elihood", "E");
        m_l7.put("encible", "A");
        m_l7.put("entally", "A");
        m_l7.put("entials", "A");
        m_l7.put("entiate", "A");
        m_l7.put("entness", "A");
        m_l7.put("fulness", "A");
        m_l7.put("ibility", "A");
        m_l7.put("icalism", "A");
        m_l7.put("icalist", "A");
        m_l7.put("icality", "A");
        m_l7.put("icalize", "A");
        m_l7.put("ication", "G");
        m_l7.put("icianry", "A");
        m_l7.put("ination", "A");
        m_l7.put("ingness", "A");
        m_l7.put("ionally", "A");
        m_l7.put("isation", "A");
        m_l7.put("ishness", "A");
        m_l7.put("istical", "A");
        m_l7.put("iteness", "A");
        m_l7.put("iveness", "A");
        m_l7.put("ivistic", "A");
        m_l7.put("ivities", "A");
        m_l7.put("ization", "F");
        m_l7.put("izement", "A");
        m_l7.put("oidally", "A");
        m_l7.put("ousness", "A");
        m_l6 = new HashMap();
        m_l6.put("aceous", "A");
        m_l6.put("acious", "B");
        m_l6.put("action", "G");
        m_l6.put("alness", "A");
        m_l6.put("ancial", "A");
        m_l6.put("ancies", "A");
        m_l6.put("ancing", "B");
        m_l6.put("ariser", "A");
        m_l6.put("arized", "A");
        m_l6.put("arizer", "A");
        m_l6.put("atable", "A");
        m_l6.put("ations", "B");
        m_l6.put("atives", "A");
        m_l6.put("eature", "Z");
        m_l6.put("efully", "A");
        m_l6.put("encies", "A");
        m_l6.put("encing", "A");
        m_l6.put("ential", "A");
        m_l6.put("enting", "C");
        m_l6.put("entist", "A");
        m_l6.put("eously", "A");
        m_l6.put("ialist", "A");
        m_l6.put("iality", "A");
        m_l6.put("ialize", "A");
        m_l6.put("ically", "A");
        m_l6.put("icance", "A");
        m_l6.put("icians", "A");
        m_l6.put("icists", "A");
        m_l6.put("ifully", "A");
        m_l6.put("ionals", "A");
        m_l6.put("ionate", "D");
        m_l6.put("ioning", "A");
        m_l6.put("ionist", "A");
        m_l6.put("iously", "A");
        m_l6.put("istics", "A");
        m_l6.put("izable", "E");
        m_l6.put("lessly", "A");
        m_l6.put("nesses", "A");
        m_l6.put("oidism", "A");
        m_l5 = new HashMap();
        m_l5.put("acies", "A");
        m_l5.put("acity", "A");
        m_l5.put("aging", "B");
        m_l5.put("aical", "A");
        if (!m_CompMode) {
            m_l5.put("alist", "A");
        }
        m_l5.put("alism", "B");
        m_l5.put("ality", "A");
        m_l5.put("alize", "A");
        m_l5.put("allic", "b");
        m_l5.put("anced", "B");
        m_l5.put("ances", "B");
        m_l5.put("antic", "C");
        m_l5.put("arial", "A");
        m_l5.put("aries", "A");
        m_l5.put("arily", "A");
        m_l5.put("arity", "B");
        m_l5.put("arize", "A");
        m_l5.put("aroid", "A");
        m_l5.put("ately", "A");
        m_l5.put("ating", "I");
        m_l5.put("ation", "B");
        m_l5.put("ative", "A");
        m_l5.put("ators", "A");
        m_l5.put("atory", "A");
        m_l5.put("ature", "E");
        m_l5.put("early", "Y");
        m_l5.put("ehood", "A");
        m_l5.put("eless", "A");
        if (!m_CompMode) {
            m_l5.put("elily", "A");
        } else {
            m_l5.put("elity", "A");
        }
        m_l5.put("ement", "A");
        m_l5.put("enced", "A");
        m_l5.put("ences", "A");
        m_l5.put("eness", "E");
        m_l5.put("ening", "E");
        m_l5.put("ental", "A");
        m_l5.put("ented", "C");
        m_l5.put("ently", "A");
        m_l5.put("fully", "A");
        m_l5.put("ially", "A");
        m_l5.put("icant", "A");
        m_l5.put("ician", "A");
        m_l5.put("icide", "A");
        m_l5.put("icism", "A");
        m_l5.put("icist", "A");
        m_l5.put("icity", "A");
        m_l5.put("idine", "I");
        m_l5.put("iedly", "A");
        m_l5.put("ihood", "A");
        m_l5.put("inate", "A");
        m_l5.put("iness", "A");
        m_l5.put("ingly", "B");
        m_l5.put("inism", "J");
        m_l5.put("inity", "c");
        m_l5.put("ional", "A");
        m_l5.put("ioned", "A");
        m_l5.put("ished", "A");
        m_l5.put("istic", "A");
        m_l5.put("ities", "A");
        m_l5.put("itous", "A");
        m_l5.put("ively", "A");
        m_l5.put("ivity", "A");
        m_l5.put("izers", "F");
        m_l5.put("izing", "F");
        m_l5.put("oidal", "A");
        m_l5.put("oides", "A");
        m_l5.put("otide", "A");
        m_l5.put("ously", "A");
        m_l4 = new HashMap();
        m_l4.put("able", "A");
        m_l4.put("ably", "A");
        m_l4.put("ages", "B");
        m_l4.put("ally", "B");
        m_l4.put("ance", "B");
        m_l4.put("ancy", "B");
        m_l4.put("ants", "B");
        m_l4.put("aric", "A");
        m_l4.put("arly", "K");
        m_l4.put("ated", "I");
        m_l4.put("ates", "A");
        m_l4.put("atic", "B");
        m_l4.put("ator", "A");
        m_l4.put("ealy", "Y");
        m_l4.put("edly", "E");
        m_l4.put("eful", "A");
        m_l4.put("eity", "A");
        m_l4.put("ence", "A");
        m_l4.put("ency", "A");
        m_l4.put("ened", "E");
        m_l4.put("enly", "E");
        m_l4.put("eous", "A");
        m_l4.put("hood", "A");
        m_l4.put("ials", "A");
        m_l4.put("ians", "A");
        m_l4.put("ible", "A");
        m_l4.put("ibly", "A");
        m_l4.put("ical", "A");
        m_l4.put("ides", "L");
        m_l4.put("iers", "A");
        m_l4.put("iful", "A");
        m_l4.put("ines", "M");
        m_l4.put("ings", "N");
        m_l4.put("ions", "B");
        m_l4.put("ious", "A");
        m_l4.put("isms", "B");
        m_l4.put("ists", "A");
        m_l4.put("itic", "H");
        m_l4.put("ized", "F");
        m_l4.put("izer", "F");
        m_l4.put("less", "A");
        m_l4.put("lily", "A");
        m_l4.put("ness", "A");
        m_l4.put("ogen", "A");
        m_l4.put("ward", "A");
        m_l4.put("wise", "A");
        m_l4.put("ying", "B");
        m_l4.put("yish", "A");
        m_l3 = new HashMap();
        m_l3.put("acy", "A");