ES ik分词器使用技巧

match查询会将查询词分词，而后对分词的结果进行term查询。

而后默认是将每一个分词term查询以后的结果求交集，因此只要分词的结果可以命中，某条数据就能够被查询出来，而分词是在新建索引时指定的，只有text类型的数据才能设置分词策略。

新建索引，并指定分词策略：

PUT mail_test3
{
  "settings": {
    "index": {
      "refresh_interval": "30s",
      "number_of_shards": "1",
      "number_of_replicas": "0"
    }
  },
  "mappings": {
    "default": {
      "_all": {
        "enabled": false
      },
      "_source": {
        "enabled": true
      },
      "properties": {
        "addressTude": {
          "type": "text",
          "analyzer": "ik_max_word",
          "search_analyzer": "ik_smart",
          "copy_to": [
            "commonText"
          ],
          "fielddata": true
        },
        "captureTime": {
          "type": "long"
        },
        "commonText": {
          "type": "text",
          "analyzer": "ik_max_word",
          "search_analyzer": "ik_smart",
          "fielddata": true
        },
        "commonNum":{
          "type": "text",
          "analyzer": "ik_max_word",
          "search_analyzer": "ik_smart",
          "fielddata": true
        },
        "imsi": {
          "type": "keyword",
          "copy_to": ["commonNum"]
        },
        "uuid": {
          "type": "keyword"
        }
      }
    }
  }
}

analyzer 指的是在建索引时的分词策略，search_analyzer 指的是在查询时的分词策略。ik分词器还有一种ik_smart 的分词策略，能够比较两种分词策略的差异：

ik_smart分词策略：

GET mail_test3/_analyze
{
  "analyzer": "ik_smart",
  "text": "湖南省湘潭市江山路96号-11-8"
}

结果：

{
  "tokens": [
    {
      "token": "湖南省",
      "start_offset": 0,
      "end_offset": 3,
      "type": "CN_WORD",
      "position": 0
    },
    {
      "token": "湘潭市",
      "start_offset": 3,
      "end_offset": 6,
      "type": "CN_WORD",
      "position": 1
    },
    {
      "token": "江",
      "start_offset": 6,
      "end_offset": 7,
      "type": "CN_CHAR",
      "position": 2
    },
    {
      "token": "山路",
      "start_offset": 7,
      "end_offset": 9,
      "type": "CN_WORD",
      "position": 3
    },
    {
      "token": "96号",
      "start_offset": 9,
      "end_offset": 12,
      "type": "TYPE_CQUAN",
      "position": 4
    },
    {
      "token": "11-8",
      "start_offset": 13,
      "end_offset": 17,
      "type": "LETTER",
      "position": 5
    }
  ]
}

ik_max_word分词策略：

GET mail_test1/_analyze
{
  "analyzer": "ik_max_word",
  "text": "湖南省湘潭市江山路96号-11-8"
}

分词结果：

{
  "tokens": [
    {
      "token": "湖南省",
      "start_offset": 0,
      "end_offset": 3,
      "type": "CN_WORD",
      "position": 0
    },
    {
      "token": "湖南",
      "start_offset": 0,
      "end_offset": 2,
      "type": "CN_WORD",
      "position": 1
    },
    {
      "token": "省",
      "start_offset": 2,
      "end_offset": 3,
      "type": "CN_CHAR",
      "position": 2
    },
    {
      "token": "湘潭市",
      "start_offset": 3,
      "end_offset": 6,
      "type": "CN_WORD",
      "position": 3
    },
    {
      "token": "湘潭",
      "start_offset": 3,
      "end_offset": 5,
      "type": "CN_WORD",
      "position": 4
    },
    {
      "token": "市",
      "start_offset": 5,
      "end_offset": 6,
      "type": "CN_CHAR",
      "position": 5
    },
    {
      "token": "江山",
      "start_offset": 6,
      "end_offset": 8,
      "type": "CN_WORD",
      "position": 6
    },
    {
      "token": "山路",
      "start_offset": 7,
      "end_offset": 9,
      "type": "CN_WORD",
      "position": 7
    },
    {
      "token": "96",
      "start_offset": 9,
      "end_offset": 11,
      "type": "ARABIC",
      "position": 8
    },
    {
      "token": "号",
      "start_offset": 11,
      "end_offset": 12,
      "type": "COUNT",
      "position": 9
    },
    {
      "token": "11-8",
      "start_offset": 13,
      "end_offset": 17,
      "type": "LETTER",
      "position": 10
    },
    {
      "token": "11",
      "start_offset": 13,
      "end_offset": 15,
      "type": "ARABIC",
      "position": 11
    },
    {
      "token": "8",
      "start_offset": 16,
      "end_offset": 17,
      "type": "ARABIC",
      "position": 12
    }
  ]
}

ik_max_word分词器的分词结果更多，分词的粒度更细，而ik_smart的分词结果粒度更粗，但较为智能。通常的策略是创建索引使用ik_max_word，查询时使用ik_smart，这样就能尽量多的查到结果，并且上文提到，match查询最终是转化为term查询，所以只要某个分词结果命中，结果中就会有该条数据。

若是对搜索结果的精度较高，能够在查询中加入operator参数，而后让分词结果的每一个term查询结果之间求交集，这样能尽量地提升精度。

这里的operator设置为or和and的差异较大，能够测试进行比较：

GET mail_test3/_search
{
  "query": {
    "match": {
      "commonText": {
         "query": "湖北省宜昌市天台东二街",
         "operator": "and"
      }
    }
  }
}