ElasticSearch-query-match-phrase短语查询
2025-01-22 08:19:30    884 字    #elasticSearch
This post is also available in English and alternative languages.

ElasticSearch版本：6.5.0(点击跳转官方文档)

短语查询


短语查询，match_phrase query 首先会把query内容分词，可以指定分词器；

同时文档还要满足两个条件

  1. 分词后所有词项都要出现在字段中
  2. 字段中词项顺序要一致

默认使用 match_phrase 会精确匹配查询的短语，需要词项、顺序完全一致（标点除外）。


1. 测试数据

新插入一批测试数据

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
POST /bookdatas/bookType
{
  "author":"cyx",
  "price":11,
  "publish":"工业出版社",
  "name":"cyx",
  "type":"大学教材",
  "info":"java is good language, and also spark is very good."
}

POST /bookdatas/bookType
{
  "author":"cyx",
  "price":11,
  "publish":"工业出版社",
  "name":"cyx",
  "type":"大学教材",
  "info":"java spark are very related, because scala is spark's programming language and scala is also based on jvm like java"
}

POST /bookdatas/bookType
{
  "author":"cyx",
  "price":11,
  "publish":"工业出版社",
  "name":"cyx",
  "type":"大学教材",
  "info":"java are spark very related, because scala is spark's programming language and scala is also based on jvm like java"
}

POST /bookdatas/bookType
{
  "author":"cyx1",
  "price":12,
  "publish":"工业出版社",
  "name":"cyx1",
  "type":"大学教材",
  "info":"spark are java very related, because scala is spark's programming language and scala is also based on jvm like java"
}

2. DSL

1
2
3
4
5
6
7
8
9
10
11
12
GET /bookdatas/bookType/_search
{
  "query": {
    "match_phrase": {
      "info": {
        "query": "java spark",
        "analyzer": "ik_max_word"
      }
    }
  },
  "_source": "info"
}

结果集

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
"hits" : [
  {
    "_index" : "bookdatas",
    "_type" : "bookType",
    "_id" : "94YAD3MBUNaKM7SA-jQU",
    "_score" : 15.382665,
    "_source" : {
      "info" : "java spark are very related, because scala is spark's programming language and scala is also based on jvm like java"
    }
  },
  {
    "_index" : "bookdatas",
    "_type" : "bookType",
    "_id" : "-IYBD3MBUNaKM7SAAzQJ",
    "_score" : 12.4879465,
    "_source" : {
      "info" : "java are spark very related, because scala is spark's programming language and scala is also based on jvm like java"
    }
  }
]

返回文档的info字段中，包含了"java"、“spark” 两个词项。

同时，它也支持通过 analyzer 指定分词器。


2.1. slop

上面说，默认使用中会精确匹配，在正常使用中比较苛刻。

就比如，第一条测试用例就检索不出来。因为该条数据中，java、spark两个词项分开的太远。

slop参数控制match_phrase查询，每个词项之间，允许相隔几个词项仍能将文档视为匹配。

第一条测试数据为例：

javaisgoodlanguageandalsoapsrk
javaspark
java→spark
java→→
java→→spark
java→→→spark
java→→→→spark
java→→→→→spark
1
2
3
4
5
6
7
8
9
10
11
12
GET /bookdatas/_search
{
  "query": {
    "match_phrase": {
      "info": {
        "query": "java spark",
          "analyzer": "ik_max_word",
          "slop": 7
      }
    }
  }
}

3. match 和 match_phrase 的区别

match query，会将输入的query拆解开，去倒排索引里面一一匹配，只要匹配上任意一个拆解后的单词，就可以作为结果返回。

当不想将搜索字符拆分，就需要使用 match_phrase 短语搜索；它会将给定的短语（phrase）当成一个完整的查询条件。

当使用match_phrase进行搜索的时候，结果集中所有的Document都必须包含你指定的查询词组。


4. Reference