elasticsearch自动补全详解

一、参考

Suggesters

Elasticsearch Suggester 详解

二、基本介绍

2.1 bing 示例

2.2 suggest 过程

三、ES 的 suggester

3.1 实现原理

将输入的文本分解为token, 然后在索引的字典中查找相似的 term 并且返回

3.2 4 种 suggester

(1) term suggester

(2) phrase suggester

(3) completion suggester

(4) context suggester

四、term suggester

(1) 创建索引，写入文档

# 创建索引
PUT yztest/
{
  "mappings": {
    "properties": {
      "message": {
        "type": "text"
      }
    }
  }
}

# 添加文档1
POST yztest/_doc/1
{
  "message": "The goal of Apache Lucene is to provide world class search capabilities"
}

# 添加文档2
POST yztest/_doc/2
{
  "message": "Lucene is the search core of both Apache Solr and Elasticsearch."
}

(2) 查看分词 token

# 分析分词器结果

GET yztest/_analyze
{
  "field": "message",
  "text": [
    "The goal of Apache Lucene is to provide world class search capabilities",
    "Lucene is the search core of both Apache Solr and Elasticsearch."
  ]
}

(3) 不同的查询结果

a) 当输入单词拼写错误时候，会推荐正确的拼写单词列表

# 查询
POST yztest/_search
{
  "suggest": {
    "suggest_message": { # 自定义的suggester名称
      "text": "lucenl",  # 查询的字符串，即用户输入的内容
      "term": { # suggester类型为term suggester
        "field": "message", # 待匹配字段
        "suggest_mode": "missing" # 推荐结果模式,missing表示如果存在了term和用户输入的文本相同，则不再推荐
      }
    }
  }
}

# 返回结果
{
  "took" : 7,
  "timed_out" : false,
  "_shards" : {
    "total" : 1,
    "successful" : 1,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : {
      "value" : 0,
      "relation" : "eq"
    },
    "max_score" : null,
    "hits" : [ ]
  },
  "suggest" : {
    "suggest_message" : [
      {
        "text" : "lucenl",
        "offset" : 0,
        "length" : 6,
        "options" : [ # options为一个数组，里面的值为具体的推荐值
          {
            "text" : "lucene",
            "score" : 0.8333333,
            "freq" : 2
          }
        ]
      }
    ]
  }
}

b) 当输入为多个单词组成的字符串时

# 查询
POST yztest/_search
{
  "suggest": {
    "suggest_message": {
      "text": "lucene search",
      "term": {
        "field": "message",
        "suggest_mode": "always"
      }
    }
  }
}

# 查询结果
{
  "took" : 3,
  "timed_out" : false,
  "_shards" : {
    "total" : 1,
    "successful" : 1,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : {
      "value" : 0,
      "relation" : "eq"
    },
    "max_score" : null,
    "hits" : [ ]
  },
  "suggest" : {
    "suggest_message" : [
      {
        "text" : "lucene",
        "offset" : 0,
        "length" : 6,
        "options" : [ ]
      },
      {
        "text" : "search",
        "offset" : 7,
        "length" : 6,
        "options" : [ ]
      }
    ]
  }
}

五、phrase suggester

# 词组查询
POST yztest/_search
{
  "suggest": {
    "YOUR_SUGGESTION": {
      "text": "Solr and Elasticearc", # 用户输入的字符串
      "phrase": { # 指定suggest类型为phrase suggester
        "field": "message", # 待匹配的字段
        "highlight": { # 可以设置高亮
          "pre_tag": "<em>",
          "post_tag": "</em>"
        }
      }
    }
  }
}

# 返回结果
{
  "took" : 9,
  "timed_out" : false,
  "_shards" : {
    "total" : 1,
    "successful" : 1,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : {
      "value" : 0,
      "relation" : "eq"
    },
    "max_score" : null,
    "hits" : [ ]
  },
  "suggest" : {
    "YOUR_SUGGESTION" : [
      {
        "text" : "Solr and Elasticearc",
        "offset" : 0,
        "length" : 20,
        "options" : [
          {
            "text" : "solr and elasticsearch",
            "highlighted" : "solr and <em>elasticsearch</em>", # 高亮部分
            "score" : 0.017689342
          }
        ]
      }
    ]
  }
}

六、completion suggester

自动补全功能

6.1 创建 mapping 指定 suggest 字段

# 创建索引
PUT yztest/
{
  "mappings": {
    "properties": {
      "message": { # 通过字段的type，指定是否使用suggest
        "type": "completion"
      }
    }
  }
}

6.2 查询

(1) 索引文档

POST yztest/_doc/1
{
  "message": "The goal of Apache Lucene is to provide world class search capabilities"
}

POST yztest/_doc/2
{
  "message": "Lucene is the search core of both Apache Solr and Elasticsearch."
}

POST yztest/_doc/3
{
  "message": "Lucene is the search core of Elasticsearch."
}

POST yztest/_doc/4
{
  "message": "Lucene is the search core of Apache Solr."
}

(2) 前缀查询

# 查询
POST yztest/_search
{
  "suggest": {
    "message_suggest": { # 自定义suggester名称
      "prefix": "lucene is the", # 前缀字符串，即用户输入的文本
      "completion": { # 指定suggester的类型为 completion suggester
        "field": "message" # 待匹配的字段
      }
    }
  }
}

# 查询结果
{
  "took" : 1,
  "timed_out" : false,
  "_shards" : {
    "total" : 1,
    "successful" : 1,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : {
      "value" : 0,
      "relation" : "eq"
    },
    "max_score" : null,
    "hits" : [ ]
  },
  "suggest" : {
    "message_suggest" : [
      {
        "text" : "lucene is the",
        "offset" : 0,
        "length" : 13,
        "options" : [
          {
            "text" : "Lucene is the search core of Apache Solr.",
            "_index" : "yztest",
            "_type" : "_doc",
            "_id" : "4",
            "_score" : 1.0,
            "_source" : {
              "message" : "Lucene is the search core of Apache Solr."
            }
          },
          {
            "text" : "Lucene is the search core of Elasticsearch.",
            "_index" : "yztest",
            "_type" : "_doc",
            "_id" : "3",
            "_score" : 1.0,
            "_source" : {
              "message" : "Lucene is the search core of Elasticsearch."
            }
          },
          {
            "text" : "Lucene is the search core of both Apache Solr and ",
            "_index" : "yztest",
            "_type" : "_doc",
            "_id" : "2",
            "_score" : 1.0,
            "_source" : {
              "message" : "Lucene is the search core of both Apache Solr and Elasticsearch."
            }
          }
        ]
      }
    ]
  }
}

(3) skip_duplicates

删除重复匹配文档

# 查询中指定skip_duplicates, 默认值为false
POST yztest/_search
{
  "suggest": {
    "message_suggest": {
      "prefix": "lucene is the",
      "completion": {
        "field": "message",
        "skip_duplicates": true
      }
    }
  }
}

(4) fuzzy query

# 查询中指定fuzzy属性，即不一定是prefix准确查询
POST yztest/_search
{
  "suggest": {
    "message_suggest": {
      "prefix": "lucen is the",
      "completion": {
        "field": "message",
        "fuzzy": {
          "fuzziness": 2
        }
      }
    }
  }
}

(5) regex 查询，正则匹配

# 正则匹配
POST yztest/_search
{
  "suggest": {
    "message_suggest": {
      "regex": ".*solr.*", # 正则表达式
      "completion": {
        "field": "message"
      }
    }
  }
}

七、context suggester

八、如何实现？