1 搜索分离

通过 Search Template实现代码调用与语句优化的分离

 POST _scripts/tmdb
 {
   "script": {
     "lang": "mustache",
     "source": {
       "_source": [
         "title","overview"
       ],
       "size": 20,
       "query": {
         "multi_match": {
           "query": "{{q}}",
           "fields": ["title","overview"]
         }
       }
     }
   }
 }
 
 POST tmdb/_search/template
 {
     "id":"tmdb",
     "params": {
         "q": "basketball with cartoon aliens"
     }
 }
 #程序传入查询条件。其他的参数配置都在_scripts/tmdb中配置
复制

2 搜索优化

• 功能

• Complete & Context Suggester

• Term & Phrase Suggester

• 搜索推荐

• 自动补全

• 比较

• Completion > Phrase > Term

• Term > Phrase > Completion

• Completion > Phrase > Term

• 精准度

• 召回率

• 性能

2.1 搜索推荐

将输入的文本分解为Token，然后在索引的字典里查找相似的Term并返回

 DELETE articles
 PUT articles
 {
   "mappings": {
     "properties": {
       "title_completion":{
         "type": "completion"
       }
     }
   }
 }
 
 POST articles/_bulk
 { "index" : { } }
 { "title_completion": "lucene is very cool"}
 { "index" : { } }
 { "title_completion": "Elasticsearch builds on top of lucene"}
 { "index" : { } }
 { "title_completion": "Elasticsearch rocks"}
 { "index" : { } }
 { "title_completion": "elastic is the company behind ELK stack"}
 { "index" : { } }
 { "title_completion": "Elk stack rocks"}
 { "index" : {} }
 
 
 POST articles/_search?pretty
 {
   "size": 0,
   "suggest": {
     "article-suggester": {
       "prefix": "elk ",
       "completion": {
         "field": "title_completion"
       }
     }
   }
 }
 
 DELETE articles
 
 POST articles/_bulk
 { "index" : { } }
 { "body": "lucene is very cool"}
 { "index" : { } }
 { "body": "Elasticsearch builds on top of lucene"}
 { "index" : { } }
 { "body": "Elasticsearch rocks"}
 { "index" : { } }
 { "body": "elastic is the company behind ELK stack"}
 { "index" : { } }
 { "body": "Elk stack rocks"}
 { "index" : {} }
 {  "body": "elasticsearch is rock solid"}
 
 
 POST _analyze
 {
   "analyzer": "standard",
   "text": ["Elk stack  rocks rock"]
 }
 
 POST /articles/_search
 {
   "size": 1,
   "query": {
     "match": {
       "body": "lucen rock"
     }
   },
   "suggest": {
     "term-suggestion": {
       "text": "lucen rock",
       "term": {
         "suggest_mode": "missing",
         "field": "body"
       }
     }
   }
 }
 #到字段body中搜索，当无法搜索到结果时(missing),返回建议的词
 #suggest_mode支持：missing，如果索引中已经存在，则不提供建议；popular 推荐出现频次更高的词；always 无论是否存在，都提供建议
 
 
 POST /articles/_search
 {
 
   "suggest": {
     "term-suggestion": {
       "text": "lucen rock",
       "term": {
         "suggest_mode": "popular",
         "field": "body"
       }
     }
   }
 }
 
 
 POST /articles/_search
 {
 
   "suggest": {
     "term-suggestion": {
       "text": "lucen rock",
       "term": {
         "suggest_mode": "always",
         "field": "body",
       }
     }
   }
 }
 
 
 POST /articles/_search
 {
 
   "suggest": {
     "term-suggestion": {
       "text": "lucen hocks",
       "term": {
         "suggest_mode": "always",
         "field": "body",
         "prefix_length":0,
         "sort": "frequency"
       }
     }
   }
 }
 #sort用来设置排序，也可以按照频率(frequency)
 #prefix_length 默认首字母不一样不会匹配推荐，但是若该参数为0则会推荐匹配
 
 POST /articles/_search
 {
   "suggest": {
     "my-suggestion": {
       "text": "lucne and elasticsear rock hello world ",
       "phrase": {
         "field": "body",
         "max_errors":2,
         "confidence":0,
         "direct_generator":[{
           "field":"body",
           "suggest_mode":"always"
         }],
         "highlight": {
           "pre_tag": "<em>",
           "post_tag": "</em>"
         }
       }
     }
   }
 }
 #Phrase Suggester相较于Term Suggester增加一些逻辑
 #max_errors 最多可以拼错的Terms
 #confidence 限制返回结果数，默认为1
复制

2.2 自动补全

Completion Suggester提供了"自动完成"的功能。用户每输入一个字符，就需要即时发送一个查询请求到后端查找匹配项

对性能要求比较苛刻。Elasticsearch采用了不同的数据结构，并非通过倒排索引来完成。而是将Analyer的数据编码成FST和索引一起存放。FST会被ES整个加载进内存。速度很快

FST只能用于前缀查找

 DELETE articles
 PUT articles
 {
   "mappings": {
     "properties": {
       "title_completion":{
         "type": "completion"
       }
     }
   }
 }
 #事先需要在mapping中设置字段类型为completion
 
 POST articles/_bulk
 { "index" : { } }
 { "title_completion": "lucene is very cool"}
 { "index" : { } }
 { "title_completion": "Elasticsearch builds on top of lucene"}
 { "index" : { } }
 { "title_completion": "Elasticsearch rocks"}
 { "index" : { } }
 { "title_completion": "elastic is the company behind ELK stack"}
 { "index" : { } }
 { "title_completion": "Elk stack rocks"}
 { "index" : {} }
 
 
 POST articles/_search?pretty
 {
   "size": 0,
   "suggest": {
     "article-suggester": {
       "prefix": "elk ",
       "completion": {
         "field": "title_completion"
       }
     }
   }
 }
 #查询时指定字段及前缀
 
 DELETE comments
 PUT comments
 PUT comments/_mapping
 {
   "properties": {
     "comment_autocomplete":{
       "type": "completion",
       "contexts":[{
         "type":"category",
         "name":"comment_category"
       }]
     }
   }
 }
 #contexts用于指定上下午，方便针对不同场景查询不同上下文。如同样是Star根据场景不同分别查询咖啡相关(Starbucks)、电影相关(star wars)
 #contexts中的类型支持category(任意字符串)、Geo(地理位置信息)
 
 POST comments/_doc
 {
   "comment":"I love the star war movies",
   "comment_autocomplete":{
     "input":["star wars"],
     "contexts":{
       "comment_category":"movies"
     }
   }
 }
 
 POST comments/_doc
 {
   "comment":"Where can I find a Starbucks",
   "comment_autocomplete":{
     "input":["starbucks"],
     "contexts":{
       "comment_category":"coffee"
     }
   }
 }
 
 
 POST comments/_search
 {
   "suggest": {
     "MY_SUGGESTION": {
       "prefix": "sta",
       "completion":{
         "field":"comment_autocomplete",
         "contexts":{
           "comment_category":"coffee"
         }
       }
     }
   }
 }
复制