lucene加权_模仿百度(Baidu)推广

我们在百度上面搜索一些东西的时候,会在最前面出现一些不是我们想要的信息

如:

在百度这里,是做了加权操作的,也就是我们知道的百度推广

那么现在我们就模拟一下百度推广

===================================================================

项目结构:

运行代码:

1 TermQuery query = new TermQuery(new Term("content","hello"));

运行效果:

也就是说,当我们搜索"hello"关键字的时候,会把所有包含关键字"hello"的结果返回出来,再依据各自的得分大小

进行结果排序,所以就算最后一个包含两个关键字"hello",但是还是排在后面,权值高的会排在前面。

运行代码:

1 TermQuery query = new TermQuery(new Term("content","ee"));

运行效果:

当我们搜索包含关键字为"ee"的时候,会出现一条记录,为什么呢?其实很好理解,因为在所有的记录中,包含关键字"ee"的

结果只有一条记录,其他记录虽然权值要比上面的结果大,但是没有包含关键字"ee",所以显示的结果也是理所当然。

声明:如果有不正确的地方,还请大家多多指教,大家可以一起探讨

================================================================

代码部分:

================================================================

/lucene_0300_boost/src/com/b510/lucene/util/LuceneUtil.java

  1 /**
  2  * 
  3  */
  4 package com.b510.lucene.util;
  5 
  6 import java.io.File;
  7 import java.io.IOException;
  8 import java.util.HashMap;
  9 import java.util.Map;
 10 
 11 import org.apache.lucene.analysis.standard.StandardAnalyzer;
 12 import org.apache.lucene.document.Document;
 13 import org.apache.lucene.document.Field;
 14 import org.apache.lucene.index.CorruptIndexException;
 15 import org.apache.lucene.index.IndexReader;
 16 import org.apache.lucene.index.IndexWriter;
 17 import org.apache.lucene.index.IndexWriterConfig;
 18 import org.apache.lucene.index.Term;
 19 import org.apache.lucene.search.IndexSearcher;
 20 import org.apache.lucene.search.ScoreDoc;
 21 import org.apache.lucene.search.TermQuery;
 22 import org.apache.lucene.search.TopDocs;
 23 import org.apache.lucene.store.Directory;
 24 import org.apache.lucene.store.FSDirectory;
 25 import org.apache.lucene.store.LockObtainFailedException;
 26 import org.apache.lucene.util.Version;
 27 
 28 /**
 29  * @author Hongten <br />
 30  * @date 2013-1-31
 31  */
 32 public class LuceneUtil {
 33 
 34     /**
 35      * 邮件id
 36      */
 37     private String[] ids = { "1", "2", "3", "4", "5", "6" };
 38     /**
 39      * 邮箱
 40      */
 41     private String[] emails = { "aa@sina.com", "bb@foxmail.com", "cc@qq.com",
 42             "dd@163.com", "ee@gmail.com", "ff@sina.com" };
 43     /**
 44      * 邮件内容
 45      */
 46     private String[] contents = { "hello,aa,hi,hell world!!", 
 47                                   "hello,bb,i'm a boy", 
 48                                   "hello,cc",
 49                                   "hello,dd,welcome to my zone,this is a test hello", 
 50                                   "hello,ee,haha,xixi,hello world!!", 
 51                                   "hello,ff" };
 52 
 53     /**
 54      * 收件人的名称
 55      */
 56     private String[] names = { "hongten", "hanyuan", "Devide", "Tom", "Steven",
 57             "Shala" };
 58 
 59     private Directory directory = null;
 60     /**
 61      * 评分
 62      */
 63     private Map<String, Float> scores = new HashMap<String, Float>();
 64     
 65     public LuceneUtil() {
 66         try {
 67             scores.put("sina.com", 1.0f);
 68             scores.put("foxmail.com", 1.1f);
 69             directory = FSDirectory.open(new File(
 70                     "D:/WordPlace/lucene/lucene_0300_boost/lucene/index"));
 71         } catch (IOException e) {
 72             e.printStackTrace();
 73         }
 74     }
 75 
 76     /**
 77      * 创建索引
 78      */
 79     public void index() {
 80         IndexWriter writer = null;
 81         try {
 82             writer = new IndexWriter(directory, new IndexWriterConfig(
 83                     Version.LUCENE_35, new StandardAnalyzer(Version.LUCENE_35)));
 84             //删除之前所建立的全部索引
 85             writer.deleteAll();
 86             // 创建文档
 87             Document document = null;
 88             for (int i = 0; i < ids.length; i++) {
 89                 // Field.Store.YES:将会存储域值,原始字符串的值会保存在索引,以此可以进行相应的回复操作,对于主键,标题可以是这种方式存储
 90                 // Field.Store.NO:不会存储域值,通常与Index.ANAYLIZED和起来使用,索引一些如文章正文等不需要恢复的文档
 91                 // ==============================
 92                 // Field.Index.ANALYZED:进行分词和索引,适用于标题,内容等
 93                 // Field.Index.NOT_ANALYZED:进行索引,但是不进行分词,如身份证号码,姓名,ID等,适用于精确搜索
 94                 // Field.Index.ANALYZED_NOT_NORMS:进行分词,但是不进行存储norms信息,这个norms中包括了创建索引的时间和权值等信息
 95                 // Field.Index.NOT_ANALYZED_NOT_NORMS:不进行分词也不进行存储norms信息(不推荐)
 96                 // Field.Index.NO:不进行分词
 97                 document = new Document();
 98                 document.add(new Field("id", ids[i], Field.Store.YES,
 99                         Field.Index.NOT_ANALYZED_NO_NORMS));
100                 document.add(new Field("email", emails[i], Field.Store.YES,
101                         Field.Index.NOT_ANALYZED));
102                 document.add(new Field("content", contents[i], Field.Store.YES,
103                         Field.Index.ANALYZED));
104                 document.add(new Field("name", names[i], Field.Store.YES,
105                         Field.Index.NOT_ANALYZED_NO_NORMS));
106                 //这里进行加权处理
107                 String et = emails[i].substring(emails[i].lastIndexOf("@")+1);
108                 System.out.println(et);
109                 if(scores.containsKey(et)){
110                     document.setBoost(scores.get(et));
111                 }else{
112                     document.setBoost(0.6f);
113                 }
114                 writer.addDocument(document);
115             }
116         } catch (CorruptIndexException e) {
117             e.printStackTrace();
118         } catch (LockObtainFailedException e) {
119             e.printStackTrace();
120         } catch (IOException e) {
121             e.printStackTrace();
122         } finally {
123             if (writer != null) {
124                 try {
125                     writer.close();
126                 } catch (CorruptIndexException e) {
127                     e.printStackTrace();
128                 } catch (IOException e) {
129                     e.printStackTrace();
130                 }
131             }
132         }
133     }
134     
135     /**
136      * 搜索
137      */
138     public void search(){
139         try {
140             IndexReader reader = IndexReader.open(directory);
141             IndexSearcher searcher = new IndexSearcher(reader);
142             TermQuery query = new TermQuery(new Term("content","hello"));
143             TopDocs tds =searcher.search(query, 10);
144             for(ScoreDoc sd : tds.scoreDocs){
145                 Document doc = searcher.doc(sd.doc);
146                 System.out.println("文档序号:["+sd.doc+"] 得分:["+sd.score+"] 邮件名称:["+doc.get("email")+"] 邮件人:["+doc.get("name")+"] 内容 : ["+doc.get("content")+"]");
147             }
148         } catch (CorruptIndexException e) {
149             e.printStackTrace();
150         } catch (IOException e) {
151             e.printStackTrace();
152         }
153     }
154 }

/lucene_0300_boost/src/com/b510/lucene/test/IndexTest.java

 1 /**
 2  * 
 3  */
 4 package com.b510.lucene.test;
 5 
 6 import org.junit.Test;
 7 
 8 import com.b510.lucene.util.LuceneUtil;
 9 
10 /**
11  * @author Hongten <br />
12  * @date 2013-1-31
13  */
14 public class IndexTest {
15 
16     @Test
17     public void testIndex(){
18         LuceneUtil util = new LuceneUtil();
19         util.index();
20     }
21     
22     @Test
23     public void testSearch(){
24         LuceneUtil util = new LuceneUtil();
25         util.search();
26     }
27     
28 }

I'm Hongten

 

posted @ 2013-02-01 12:01  Hongten  阅读(2721)  评论(0编辑  收藏  举报
Fork me on GitHub