elasticsearch suggest 的几种使用-completion 的基本使用

最新推荐文章于 2025-07-22 20:14:00 发布

转载最新推荐文章于 2025-07-22 20:14:00 发布 · 9.1k 阅读

elasticsearch 专栏收录该内容

49 篇文章

订阅专栏

本文对比并实践了在Solr和Elasticsearch中使用拼写建议的功能，包括TermSuggestionBuilder和PhraseSuggestionBuilder的使用方法，并详细介绍了Lucene和Elasticsearch中suggester的不同实现及映射方式。通过代码示例展示了如何在Elasticsearch中创建带有suggester的映射、添加索引时添加提示以及获取完成建议的方法。

我们再使用solr时，可以用spellchecker进行搜索时建议。在elasticsearch中我们使用SuggestionBuilder进行拼写建议。
1.使用TermSuggestionBuilder

TermSuggestionBuilder suggest=new TermSuggestionBuilder("my-suggest-1");
suggest.analyzer("ik").field("content").size(n).text(q).maxTermFreq(10).minDocFreq(0).suggestMode("always");
Client client = ESServiceSingleton2.getTransportClient();
SearchResponse response = client.prepareSearch("cms")
.setTypes("news").setSearchType(SearchType.COUNT)
.setSuggestText(q)
.addSuggestion(suggest)
.execute().actionGet();
Suggest suggest=response.getSuggest();
List<? extends Entry<? extends Option>> list=suggest.getSuggestion("simple").getEntries();
for(int i=0; i<list.size();i++){

List<?> options=list.get(i).getOptions();
for(int j=0;j<options.size();j++){
if(options.get(j) instanceof Option){
Option op=(Option)options.get(j);
System.out.println(op.getScore());
System.out.println(op.getText());

}

}
System.out.println(list.get(i).getText());
}

1.使用PhraseSuggestionBuilder

//q为查询词
//n为返回建议词条长度
Suggest suggest = searchSuggest(client, q,
phraseSuggestion("simple").maxErrors(0.5f).field("content").analyzer("ik")
.addCandidateGenerator(PhraseSuggestionBuilder.candidateGenerator("content").minWordLength(1).suggestMode("always"))
.size(n));

SearchResponse response = client.prepareSearch("cms")
.setTypes("news").setSearchType(SearchType.COUNT)
.setSuggestText(q)
.addSuggestion(suggest)
.execute().actionGet();
Suggest suggest=response.getSuggest();
List<? extends Entry<? extends Option>> list=suggest.getSuggestion("simple").getEntries();
for(int i=0; i<list.size();i++){

List<?> options=list.get(i).getOptions();
for(int j=0;j<options.size();j++){
if(options.get(j) instanceof Option){
Option op=(Option)options.get(j);
System.out.println(op.getScore());
System.out.println(op.getText());

}

}
System.out.println(list.get(i).getText());
}

在lucene里面，suggest 的支持非常完善，可以随心所欲的定制；但是在es中使用起来就没有那么方便了。 es给suggest 分类4类：term ；phrase； completion； context；目前最新版本是es1.2.1 这一块也还在不断完善中； term suggester 是根据某词元在制定的字段中出现的频次来做出提示； phrase suggester 是给term 上做了加强；这里不做使用方法介绍；

链接地址：http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/search-suggesters-term.html

剩下的是completion 、context suggester。这两个的使用与上面个使用方法完全不一样，上面都在查询的时候根据制定字段内容来做提示，而这两种是需要在mapping 里面定制suggester字段。使用时完全匹配时提示；

completion suggester 官方文档 http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/search-suggesters-completion.html 文档讲解的非常全面，遗憾的是没有代码，这里把代码加上java client；

1：设置mapping 中的suggester

@Test 
public void createQucikIndecs() throws IOException{ 
XContentBuilder mapping = XContentFactory.jsonBuilder() 
.startObject() 
.startObject("pingzhuang") 
.startObject("properties") 
.startObject("id").field("type", "long").field("store", "yes").field("index", "not_analyzed").endObject() 
.startObject("name").field("type", "string").field("store", "no").field("indexAnalyzer", "ik").field("searchAnalyzer", "ik").endObject() 
.startObject("suggest").field("type","completion").field("index_analyzer","simple").field("search_analyzer","simple").field("payloads","true").endObject() 
.endObject() 
.endObject() 
.endObject(); 
ESHandler.createQuickIndices("chinamedic", "pingzhuang", mapping, "formedic"); 
}

2：设置好了以后，在做索引的时候每次都应该添加提示

@Test 
public void addIndex(){ 
String json="{\"id\":44,\"name\":\"老狼-白衣飘飘的年代\",\"suggest\":{\"input\":\"老狼-白衣飘飘的年代\"}}"; 
String json1="{\"id\":42,\"name\":\"老狼-郎心似铁\",\"suggest\":{\"input\":\"老狼-郎心似铁\"}}"; 
String json2="{\"id\":43,\"name\":\"老狼-流浪歌手的请人\",\"suggest\":{\"input\":\"老狼-流浪歌手的请人\"}}"; 
ESHandler.addOneIndex("chinamedic","pingzhuang", json); 
ESHandler.addOneIndex("chinamedic","pingzhuang", json1); 
ESHandler.addOneIndex("chinamedic","pingzhuang", json2); 
System.out.println(); 
}

3：然后就是查询获得提示了

/** 
* 搜索建议，自动补全搜索结结果 
* @param indices 索引库名称 
* @param prefix 搜索前缀词 
* @return 建议列表 
*/ 
public static List<String> getCompletionSuggest(String indices, 
String prefix) { 
CompletionSuggestionBuilder suggestionsBuilder = new CompletionSuggestionBuilder( 
"complete"); 
suggestionsBuilder.text(prefix); 
suggestionsBuilder.field("suggest"); 
suggestionsBuilder.size(10); 
SuggestResponse resp = client.prepareSuggest(indices) 
.addSuggestion(suggestionsBuilder).execute().actionGet(); 
List<? extends Entry<? extends Option>> list = resp.getSuggest() 
.getSuggestion("complete").getEntries(); 
List<String> suggests = new ArrayList<String>(); 
if (list == null) { 
return null; 
} else { 
for (Entry<? extends Option> e : list) { 
for (Option option : e) { 
suggests.add(option.getText().toString()); 
} 
} 
return suggests; 
} 
}