欢迎您访问程序员文章站本站旨在为大家提供分享程序员计算机编程知识!
您现在的位置是: 首页

Lucene4.3进阶开发之纯阳无极(十九)

程序员文章站 2022-05-17 09:30:51
...
原创不易,转载请务必注明,原创地址,谢谢配合!
http://qindongliang.iteye.com/blog/2164583

Lucene内置很多的分词器工具包,几乎涵盖了全球所有的国家和地区,最近散仙,在搞多语言分词的一个处理,主要国家有西班牙,葡萄牙,德语,法语,意大利,其实这些语系都与英语非常类似,都是以空格为分割的语种。


那么首先,探讨下分词器的词形还原和词干提取的对搜索的意义?在这之前,先看下两者的概念:
词形还原(lemmatization),是把一个任何形式的语言词汇还原为一般形式(能表达完整语义),而词干提取

(stemming)是抽取词的词干或词根形式(不一定能够表达完整语义)。词形还原和词干提取是词形规范化的两类
重要方式,都能够达到有效归并词形的目的,二者既有联系也有区别

详细介绍,请参考这篇文章


在电商搜索里,词干的抽取,和单复数的还原比较重要(这里主要针对名词来讲),因为这有关搜索的查准率,和查全率的命中,如果我们的分词器没有对这些词做过处理,会造成什么影响呢?那么请看如下的一个例子?

句子: i have two cats

分词器如果什么都没有做:

这时候我们搜cat,就会无命中结果,而必须搜cats才能命中到一条数据,而事实上cat和cats是同一个东西,只不过单词的形式不一样,这样以来,如果不做处理,我们的查全率和查全率都会下降,会涉及影响到我们的搜索体验,所以stemming这一步,在某些场合的分词中至关重要。

本篇,散仙,会参考源码分析一下,关于德语分词中中如何做的词干提取,先看下德语的分词声明:

	 List<String> list=new ArrayList<String>();
		list.add("player");//这里面的词,不会被做词干抽取,词形还原
		CharArraySet ar=new CharArraySet(Version.LUCENE_43,list , true);
		//分词器的第二个参数是禁用词参数,第三个参数是排除不做词形转换,或单复数的词
		GermanAnalyzer sa=new GermanAnalyzer(Version.LUCENE_43,null,ar);


接着,我们具体看下,在德语的分词器中,都经过了哪几部分的过滤处理:
  protected TokenStreamComponents createComponents(String fieldName,
      Reader reader) {
	  //标准分词器过滤
    final Tokenizer source = new StandardTokenizer(matchVersion, reader);
    TokenStream result = new StandardFilter(matchVersion, source);
	//转小写过滤
    result = new LowerCaseFilter(matchVersion, result);
	//禁用词过滤
    result = new StopFilter( matchVersion, result, stopwords);
	//排除词过滤
    result = new SetKeywordMarkerFilter(result, exclusionSet);
    if (matchVersion.onOrAfter(Version.LUCENE_36)) {
	//在lucene3.6以后的版本,采用如下filter过滤
	  //规格化,将德语中的特殊字符,映射成英语
      result = new GermanNormalizationFilter(result);
	  //stem词干抽取,词性还原
      result = new GermanLightStemFilter(result);
    } else if (matchVersion.onOrAfter(Version.LUCENE_31)) {
	//在lucene3.1至3.6的版本中,采用SnowballFilter处理
      result = new SnowballFilter(result, new German2Stemmer());
    } else {
	//在lucene3.1之前的采用兼容的GermanStemFilter处理
      result = new GermanStemFilter(result);
    }
    return new TokenStreamComponents(source, result);
  }


OK,我们从源码中得知,在Lucene4.x中对德语的分词也做了向前和向后兼容,现在我们主要关注在lucene4.x之后的版本如何的词形转换,下面分别看下
     result = new GermanNormalizationFilter(result);
      result = new GermanLightStemFilter(result);
这两个类的功能:

package org.apache.lucene.analysis.de;

/*
 * Licensed to the Apache Software Foundation (ASF) under one or more
 * contributor license agreements.  See the NOTICE file distributed with
 * this work for additional information regarding copyright ownership.
 * The ASF licenses this file to You under the Apache License, Version 2.0
 * (the "License"); you may not use this file except in compliance with
 * the License.  You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

import java.io.IOException;

import org.apache.lucene.analysis.TokenFilter;
import org.apache.lucene.analysis.TokenStream;
import org.apache.lucene.analysis.tokenattributes.CharTermAttribute;
import org.apache.lucene.analysis.util.StemmerUtil;

/**
 * Normalizes German characters according to the heuristics
 * of the <a href="http://snowball.tartarus.org/algorithms/german2/stemmer.html">
 * German2 snowball algorithm</a>.
 * It allows for the fact that ä, ö and ü are sometimes written as ae, oe and ue.
 * 
 * [list]
 *   <li> 'ß' is replaced by 'ss'
 *   <li> 'ä', 'ö', 'ü' are replaced by 'a', 'o', 'u', respectively.
 *   <li> 'ae' and 'oe' are replaced by 'a', and 'o', respectively.
 *   <li> 'ue' is replaced by 'u', when not following a vowel or q.
 * [/list]
 * <p>
 * This is useful if you want this normalization without using
 * the German2 stemmer, or perhaps no stemming at all.
 *上面的解释说得很清楚,主要是对德文的一些特殊字母,转换成对应的英文处理
 *
 */
 
public final class GermanNormalizationFilter extends TokenFilter {
  // FSM with 3 states:
  private static final int N = 0; /* ordinary state */
  private static final int V = 1; /* stops 'u' from entering umlaut state */
  private static final int U = 2; /* umlaut state, allows e-deletion */

  private final CharTermAttribute termAtt = addAttribute(CharTermAttribute.class);
  
  public GermanNormalizationFilter(TokenStream input) {
    super(input);
  }

  @Override
  public boolean incrementToken() throws IOException {
    if (input.incrementToken()) {
      int state = N;
      char buffer[] = termAtt.buffer();
      int length = termAtt.length();
      for (int i = 0; i < length; i++) {
        final char c = buffer[i];
        switch(c) {
          case 'a':
          case 'o':
            state = U;
            break;
          case 'u':
            state = (state == N) ? U : V;
            break;
          case 'e':
            if (state == U)
              length = StemmerUtil.delete(buffer, i--, length);
            state = V;
            break;
          case 'i':
          case 'q':
          case 'y':
            state = V;
            break;
          case 'ä':
            buffer[i] = 'a';
            state = V;
            break;
          case 'ö':
            buffer[i] = 'o';
            state = V;
            break;
          case 'ü': 
            buffer[i] = 'u';
            state = V;
            break;
          case 'ß':
            buffer[i++] = 's';
            buffer = termAtt.resizeBuffer(1+length);
            if (i < length)
              System.arraycopy(buffer, i, buffer, i+1, (length-i));
            buffer[i] = 's';
            length++;
            state = N;
            break;
          default:
            state = N;
        }
      }
      termAtt.setLength(length);
      return true;
    } else {
      return false;
    }
  }
}

package org.apache.lucene.analysis.de;

/*
 * Licensed to the Apache Software Foundation (ASF) under one or more
 * contributor license agreements.  See the NOTICE file distributed with
 * this work for additional information regarding copyright ownership.
 * The ASF licenses this file to You under the Apache License, Version 2.0
 * (the "License"); you may not use this file except in compliance with
 * the License.  You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

import java.io.IOException;

import org.apache.lucene.analysis.TokenFilter;
import org.apache.lucene.analysis.TokenStream;
import org.apache.lucene.analysis.miscellaneous.SetKeywordMarkerFilter;
import org.apache.lucene.analysis.tokenattributes.CharTermAttribute;
import org.apache.lucene.analysis.tokenattributes.KeywordAttribute;

/**
 * A {@link TokenFilter} that applies {@link GermanLightStemmer} to stem German
 * words.
 * <p>
 * To prevent terms from being stemmed use an instance of
 * {@link SetKeywordMarkerFilter} or a custom {@link TokenFilter} that sets
 * the {@link KeywordAttribute} before this {@link TokenStream}.
 * 

 *
 *
 *这个类,主要做Stemmer(词干提取),而我们主要关注
 *GermanLightStemmer这个类的作用
 *
 *
 */
public final class GermanLightStemFilter extends TokenFilter {
  private final GermanLightStemmer stemmer = new GermanLightStemmer();
  private final CharTermAttribute termAtt = addAttribute(CharTermAttribute.class);
  private final KeywordAttribute keywordAttr = addAttribute(KeywordAttribute.class);

  public GermanLightStemFilter(TokenStream input) {
    super(input);
  }
  
  @Override
  public boolean incrementToken() throws IOException {
    if (input.incrementToken()) {
      if (!keywordAttr.isKeyword()) {
        final int newlen = stemmer.stem(termAtt.buffer(), termAtt.length());
        termAtt.setLength(newlen);
      }
      return true;
    } else {
      return false;
    }
  }
}

下面看下,在GermanLightStemmer中,如何做的词干提取:源码如下:
 package org.apache.lucene.analysis.de;

/*
 * Licensed to the Apache Software Foundation (ASF) under one or more
 * contributor license agreements.  See the NOTICE file distributed with
 * this work for additional information regarding copyright ownership.
 * The ASF licenses this file to You under the Apache License, Version 2.0
 * (the "License"); you may not use this file except in compliance with
 * the License.  You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

/* 
 * This algorithm is updated based on code located at:
 * http://members.unine.ch/jacques.savoy/clef/
 * 
 * Full copyright for that code follows:
 */

/*
 * Copyright (c) 2005, Jacques Savoy
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without 
 * modification, are permitted provided that the following conditions are met:
 *
 * Redistributions of source code must retain the above copyright notice, this 
 * list of conditions and the following disclaimer. Redistributions in binary 
 * form must reproduce the above copyright notice, this list of conditions and
 * the following disclaimer in the documentation and/or other materials 
 * provided with the distribution. Neither the name of the author nor the names 
 * of its contributors may be used to endorse or promote products derived from 
 * this software without specific prior written permission.
 * 
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" 
 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 
 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE 
 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 
 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 
 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 
 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 
 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 
 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
 * POSSIBILITY OF SUCH DAMAGE.
 */

/**
 * Light Stemmer for German.
 * <p>
 * This stemmer implements the "UniNE" algorithm in:
 * <i>Light Stemming Approaches for the French, Portuguese, German and Hungarian Languages</i>
 * Jacques Savoy
 */
public class GermanLightStemmer {
  
  //处理特殊字符映射
  public int stem(char s[], int len) {   
    for (int i = 0; i < len; i++)
      switch(s[i]) {
        case 'ä':
        case 'à':
        case 'á':
        case 'â': s[i] = 'a'; break;
        case 'ö':
        case 'ò':
        case 'ó':
        case 'ô': s[i] = 'o'; break;
        case 'ï':
        case 'ì':
        case 'í':
        case 'î': s[i] = 'i'; break;
        case 'ü': 
        case 'ù': 
        case 'ú':
        case 'û': s[i] = 'u'; break;
      }
    
    len = step1(s, len);
    return step2(s, len);
  }
  
  
  private boolean stEnding(char ch) {
    switch(ch) {
      case 'b':
      case 'd':
      case 'f':
      case 'g':
      case 'h':
      case 'k':
      case 'l':
      case 'm':
      case 'n':
      case 't': return true;
      default: return false;
    }
  }
  //处理基于以下规则的词干抽取和缩减
  private int step1(char s[], int len) {
    if (len > 5 && s[len-3] == 'e' && s[len-2] == 'r' && s[len-1] == 'n')
      return len - 3;
    
    if (len > 4 && s[len-2] == 'e')
      switch(s[len-1]) {
        case 'm':
        case 'n':
        case 'r':
        case 's': return len - 2;
      }
    
    if (len > 3 && s[len-1] == 'e')
      return len - 1;
    
    if (len > 3 && s[len-1] == 's' && stEnding(s[len-2]))
      return len - 1;
    
    return len;
  }
  //处理基于以下规则est,er,en等的词干抽取和缩减
  private int step2(char s[], int len) {
    if (len > 5 && s[len-3] == 'e' && s[len-2] == 's' && s[len-1] == 't')
      return len - 3;
    
    if (len > 4 && s[len-2] == 'e' && (s[len-1] == 'r' || s[len-1] == 'n'))
      return len - 2;
    
    if (len > 4 && s[len-2] == 's' && s[len-1] == 't' && stEnding(s[len-3]))
      return len - 2;
    
    return len;
  }
}

具体的分析结果如下:
搜索技术交流群:324714439
大数据hadoop交流群:376932160

0,将一些德语特殊字符,替换成对应的英文表示
1,将所有词干元音还原 a ,o,i,u
ste(2)(按先后顺序,符合以下任意一项,就完成一次校验(return))
2,单词长度大于5的词,以ern结尾的,直接去掉
3,单词长度大于4的词,以em,en,es,er结尾的,直接去掉
4,单词长度大于3的词,以e结尾的直接去掉
5,单词长度大于3的词,以bs,ds,fs,gs,hs,ks,ls,ms,ns,ts结尾的,直接去掉s
step(3)(按先后顺序,符合以下任意一项,就完成一次校验(return))
6,单词长度大于5的词,以est结尾的,直接去掉
7,单词长度大于4的词,以er或en结尾的直接去掉
8,单词长度大于4的词,bst,dst,fst,gst,hst,kst,lst,mst,nst,tst,直接去掉后两位字母st

最后,结合网上资料分析,基于er,en,e,s结尾的是做单复数转换的,其他的几条规则主要是对非名词的单词,做词干抽取。


原创不易,转载请务必注明,原创地址,谢谢配合!
http://qindongliang.iteye.com/blog/2164583