elasticsearch的智能搜索建议suggester

我们在使用搜索栏的时候，会自动补全或者纠错

这样的功能在es中可以使用Suggesters API。
Elasticsearch里设计了4种类别的Suggester

Term Suggester

准备索引库创建索引

PUT /chatting/
{
  "mappings": {
    "properties": {
      "body": {
        "type": "text"
      }
    }
  }
}

POST _bulk/?refresh=true
{"index":{"_index":"chatting"}}
{"body":"I am iron man"}
{"index":{"_index":"chatting"}}
{"body":"php is the best language in the world"}
{"index":{"_index":"chatting"}}
{"body":"hello world"}
{"index":{"_index":"chatting"}}
{"body":"how to speaking this word"}

分词器会把这些文档分词，放进词典
执行suggester

POST /chatting/_search
{
  "suggest": {
    "my-suggestion": {
      "text": "hallo word",
      "term": {
        "suggest_mode": "missing",
        "field": "body"
      }
    }
  }
}

如图上面hallo没有在词典中就有建议，而word在词典中能找到就没有建议，因为suggest mode是missing
然后把suggest_mode改成missing

这里因为world在词典中的词频更高，所以这里建议就变成了world
最后吧mode改成alway

这里表示总是会给出相似项
实验过后感觉这种搜索建议对中文的支持比较差，用的ik分词器，不知道有没有其他的方法

Phrase Suggester

Phrase suggester会计算词条是否同时出现在索引文档里面，相邻程度和词频等。

POST /chatting/_search
{
  "suggest": {
    "my-suggestion": {
      "text": "php is the best languages",
      "phrase": {
        "field": "body",
        "highlight": {
          "pre_tag": "<em>",
          "post_tag": "</em>"
        }
      }
    }
  }
}

被替换的词条会高亮

Completion Suggester

索引并非通过倒排来完成，而是将analyze过的数据编码成FST和索引一起存放。对于一个open状态的索引，FST会被ES整个装载到内存里的，进行前缀查找速度极快。但是FST只能用于前缀查找，这也是 Completion Suggester的局限所在。
重新定义一下索引库，造数据

PUT /chatting_completion/
{
  "mappings": {
    "properties": {
      "body": {
        "type": "completion"
      }
    }
  }
}
POST _bulk/?refresh=true
{"index":{"_index":"chatting_completion"}}
{"body":"I am iron man"}
{"index":{"_index":"chatting_completion"}}
{"body":"php is the best language in the world"}
{"index":{"_index":"chatting_completion"}}
{"body":"hello world"}
{"index":{"_index":"chatting_completion"}}
{"body":"how to speaking this word"}

前缀查询

POST /chatting_completion/_search?pretty
{
  "size": 0,
  "suggest": {
    "blog-suggest": {
      "prefix": "i am",
      "completion": {
        "field": "body"
      }
    }
  }
}

Context Suggester

Context Suggester是Completion Suggester的扩展
可以根据上下文来提供建议值
有两种类型的 Context

1. Category - 任意的字符串；
2. Geo - 地理位置信息；

创建索引，造数据
这里comments库定义了属性comment和comment_autocomplete，comment_autocomplete的type是completion并且有一个Category类型上下文名字叫comment_category

PUT comments/
{
  "mappings": {
    "properties": {
      "comment": {
        "type": "text"
      },
      "comment_autocomplete": {
        "type": "completion",
        "contexts": [
          {
            "type": "category",
            "name": "comment_category"
          }
        ]
      }
    }
  }
}


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"
    }
  }
}

查找

POST comments/_search
{
  "suggest": {
    "MY_SUGGESTION": {
      "prefix": "sta",
      "completion": {
        "field": "comment_autocomplete",
        "contexts": {
          "comment_category": "coffee"
        }
      }
    }
  }
}

这里sta在两个文档里面都有，但是查询中指定了上下文在”coffee“里面找，所以查出来是Where can I find a Starbucks