中文文本摘要生成

中文自动文摘，基于jieba分词，全Java代码。给定文本输出自定义长度的文摘。

package abstracts;

import java.util.ArrayList;
import java.util.Collections;
import java.util.linkedList;
import java.util.List;

import org.apache.commons.lang3.StringUtils;

import abstracts.fileOperation.Sentence;
import abstracts.fileOperation.SentenceComparator;
import abstracts.fileOperation.Word;

import com.huaban.analysis.jieba.Jiebasegmenter;
import com.huaban.analysis.jieba.Jiebasegmenter.SegMode;
import com.huaban.analysis.jieba.SegToken;

public class AbstractEntrance {

	private ArrayList word; // 名词动词统计
	private linkedList sentence; // 句子统计
	private String current_sentence = ““; // 当前句子
	private String current_sentence1 = ““; // 当前句子分词后结果
	private String abstracts = ““; // 文摘
	
	String[] terminators = new String[] {“。“ “？“ “！“ “……

 属性            大小     日期    时间   名称
----------- ---------  ---------- -----  ----
     目录           0  2014-08-30 16:11  abstracts-jieba\
     文件         368  2014-07-11 21:02  abstracts-jieba\.classpath
     文件         385  2014-07-11 19:34  abstracts-jieba\.project
     目录           0  2014-08-30 16:11  abstracts-jieba\.settings\
     文件         128  2014-07-11 19:36  abstracts-jieba\.settings\org.eclipse.core.resources.prefs
     文件         598  2014-07-11 19:34  abstracts-jieba\.settings\org.eclipse.jdt.core.prefs
     目录           0  2014-10-13 18:25  abstracts-jieba\bin\
     目录           0  2014-10-13 18:25  abstracts-jieba\bin\abstracts\
     文件        5769  2014-10-29 10:02  abstracts-jieba\bin\abstracts\AbstractEntrance.class
     文件       22422  2014-10-29 10:02  abstracts-jieba\bin\abstracts\AbstractMain.class
     目录           0  2014-10-13 18:25  abstracts-jieba\bin\abstracts\fileOperation\
     文件        1546  2014-10-13 18:25  abstracts-jieba\bin\abstracts\fileOperation\Sentence.class
     文件         958  2014-10-13 18:25  abstracts-jieba\bin\abstracts\fileOperation\SentenceComparator.class
     文件         837  2014-10-13 18:25  abstracts-jieba\bin\abstracts\fileOperation\Word.class
     目录           0  2014-10-13 18:25  abstracts-jieba\bin\com\
     目录           0  2014-10-13 18:25  abstracts-jieba\bin\com\huaban\
     目录           0  2014-10-13 18:25  abstracts-jieba\bin\com\huaban\analysis\
     目录           0  2014-10-13 18:25  abstracts-jieba\bin\com\huaban\analysis\jieba\
     文件        1546  2014-10-13 18:25  abstracts-jieba\bin\com\huaban\analysis\jieba\CharacterUtil.class
     文件        1222  2014-10-13 18:25  abstracts-jieba\bin\com\huaban\analysis\jieba\Jiebasegmenter$SegMode.class
     文件        8663  2014-10-13 18:25  abstracts-jieba\bin\com\huaban\analysis\jieba\Jiebasegmenter.class
     文件        8416  2014-10-13 18:25  abstracts-jieba\bin\com\huaban\analysis\jieba\JiebasegmenterTest.class
     文件        1104  2014-10-13 18:25  abstracts-jieba\bin\com\huaban\analysis\jieba\Pair.class
     文件        1162  2014-10-13 18:25  abstracts-jieba\bin\com\huaban\analysis\jieba\SegToken.class
     文件         650  2014-10-13 18:25  abstracts-jieba\bin\com\huaban\analysis\jieba\TrieNode.class
     目录           0  2014-10-13 18:25  abstracts-jieba\bin\com\huaban\analysis\jieba\viterbi\
     文件        8827  2014-10-13 18:25  abstracts-jieba\bin\com\huaban\analysis\jieba\viterbi\FinalSeg.class
     文件        2924  2014-10-13 18:25  abstracts-jieba\bin\com\huaban\analysis\jieba\Word.class
     文件        7487  2014-10-13 18:25  abstracts-jieba\bin\com\huaban\analysis\jieba\WordDictionary.class
     文件     5071839  2014-06-17 16:15  abstracts-jieba\bin\dict.txt
     文件      665440  2014-06-17 16:15  abstracts-jieba\bin\prob_emit.txt
............此处省略26个文件信息