ES搜索推荐

ES针对不同的应用场景，把suggester主要分为以下四种：

• term suggester
• phrase suggester
• completion suggester
• context suggester

Suggester examples | Elastic Documentation

term suggester

针对单独term的搜索推荐，不考虑搜索短语中多个term的关系。自动修正用户输入的拼写错误，当用户输入的词项不存在于索引中时提供相近的替代词项，处理因大小写、拼写错误等导致搜索无结果的问题。

工作原理：

1. 分词处理：将用户输入的文本拆分为单个词项
2. 生成候选词：对每个词项，从索引中查找编辑距离相近的词项
3. 排序与过滤：基于词项在索引中的频率与编辑距离对候选词进行排序，过滤掉不满足条件的候选词

参数说明

参数	说明
text	用户搜索的文本
field	要从哪个字段选取推荐数据
size	每个建议返回的最大结果数
sort	如何按照提示词项排序 * score 分数>词频>词项本身 * frequency 词频>分数>词项本身
suggest_mode	搜索推荐的推荐模式 * missing 默认，仅为不在索引中的词项生成建议词 * popular 原词存在时仅当候选词频率更高时返回建议 * always 始终返回建议词
max_edits	允许的最大编辑距离
prefix_length	前缀匹配时必须满足的最少字符
min_word_length	最少包含的单词数量
min_doc_freq	候选词的最小文档频率，过滤低频词
max_term_freq	原词的最大文档频率（高于此值的原词不生成建议）

POST /my_index/_search
{
  "suggest": {
    "my_term_suggestion": {
      "text": "musi gare",
      "term": {
        "field": "text",
        "suggest_mode": "popular",
        "size": 3,
        "max_edits": 2
      }
    }
  }
}

phrase suggester

term suggester可以对单个term进行建议或纠错，不会考虑多个term之间的关系，但是phrase suggester在term suggester的基础上，考虑多个term之间的关系

使用phrase sugges需要创建特殊的mapping：

PUT test
{
  "settings": {
    "index": {
      "number_of_shards": 1,
      "number_of_replicas": 0,
      "analysis": {
        "analyzer": {
          "trigram": {
            "type": "custom",
            "tokenizer": "standard",
            "filter": [
              "lowercase",
              "shingle"
            ]
          }
        },
        "filter": {
          "single": {
            "type": "shingle",
            "min_single_size": 2,
            "max_single_size": 3
          }
        }
      }
    }
  },
  "mappings": {
    "properties": {
      "title": {
        "type": "text",
        "fields": {
          "trigram": {
            "type": "text",
            "analyzer": "trigram"
          }
        }
      }
    }
  }
}

shingle：在单词粒度进行ngram

GET test/_search
{
  "suggest": {
    "text": "Luceen and elasticsearchc",
    "simple_phrase": {
      "phrase": {
        "field": "title.trigram",
        "max_errors": 1,
        "confidence": 1,
        "direct_generator": [
          {
            "field": "title.trigram",
            "suggest_mode": "always"
          }
        ],
        "highlight": {
          "pre_tag": "<em>",
          "post_tag": "</em>"
        }
      }
    }
  }
}

推荐的结果并不一定是真实存在的文档

completion suggester

基于内存（使用FST数据结构）而非索引，性能强悍。需要结合特定的completion类型，只适合前缀推荐。

PUT suggest_carinfo
{
  "mappings": {
    "properties": {
      "title": {
        "type": "text",
        "analyzer": "ik_max_word",
        "fields": {
          "suggest": {
            "type": "completion",
            "analyzer": "ik_max_word"
          }
        }
      },
      "content": {
        "type": "text",
        "analyzer": "ik_max_word"
      }
    }
  }
}

GET suggest_carinfo/_search
{
  "suggest": {
    "car_suggest": {
      "prefix": "宝马5系",
      "completion": {
        "field": "title.suggest",
        "skip_duplicates": true,
        "fuzzy": {
          "fuzziness": 2
        }
      }
    }
  }
}

context suggester

Context Suggester 通过在索引阶段将上下文信息与建议词条一起存储，然后在查询阶段使用这些上下文来过滤或影响建议的排序。

context suggester用法示例：

// 定义索引
PUT place
{
  "mappings": {
    "properties": {
      "suggest": {
        "type": "completion",
        "contexts": [
          {
            "name": "place_type",
            "type": "category"
          },
          {
            "name": "location",
            "type": "geo",
            "precision": 4
          }
        ]
      }
    }
  }
}

// 插入数据
PUT place/_doc/1
{
  "suggest": {
    "input": ["timmy's", "starbucks", "bunkin donuts"],
    "contexts": {
      "place_type": ["cafe", "food"]
    }
  }
}
PUT place/_doc/2
{
  "suggest": {
    "input": ["monkey", "timmy's", "Lamborghini"],
    "contexts": {
      "place_type": ["money"]
    }
  }
}
PUT place/_doc/3
{
  "suggest":{
    "input": "timmy's",
    "contexts": {
      "location": [
      {"lat": 43.6624803, "lon": -79.3863353},
      {"lat": 43.6624718, "lon": -79.3873227}
      ]
    }
  }
}

// 检索
POST place/_search?pretty
{
  "suggest": {
    "place_suggestion": {
      "prefix": "sta",
      "completion": {
        "field": "suggest",
        "size": 10,
        "contexts": {
          "place_type": ["cafe", "restaurants"]
        }
      }
    }
  }
}
POST place/_search?pretty
{
  "suggest": {
    "place_suggestion": {
      "prefix": "tim",
      "completion": {
        "field": "suggest",
        "contexts": {
          "place_type": [
          {"context": "cafe"},
          {"context": "money", "boost": 2}  // 权重提升
          ]
        }
      }
    }
  }
}
POST place/_search
{
  "suggest": {
    "place_suggestion": {
      "prefix": "tim",
      "completion": {
        "field": "suggest",
        "size": 10,
        "contexts": {
          "location": {"lat": 43.662, "lon": -79.380}
        }
      }
    }
  }
}

通过字段指定上下文：

PUT place_path_category
{
  "mappings": {
    "properties": {
      "suggest": {
        "type": "completion",
        "contexts": [
          {
            "name": "place_type",
            "type": "category",
            "path": "cat"  // 类别上下文从cat字段读取
          },
          {
            "name": "location",
            "type": "geo",
            "precision": 4,
            "path": "loc"  // 地理上下文从loc字段读取
          }
        ]
      },
      "loc": {
        "type": "geo_point"
      }
    }
  }
}

PUT place_path_category/_doc/1
{
  "suggest": ["timmy's", "starbucks", "dunkin donuts"],
  "cat": ["cafe", "food"]
}

POST place_path_category/_search?pretty
{
  "suggest": {
    "place_suggestion": {
      "prefix": "tim",
      "completion": {
        "field": "suggest",
        "size": 10,
        "contexts": {
          "place_type": ["cafe"]
        }
      }
    }
  }
}