php - Elasticsearch-如何在查询中使用多个分析器

Question

我想在查询中实现同义词和停用词过滤器。为此，我创建了两个分析器，并且两个分析器都可以单独正常工作。但是我要同时使用它们，怎么办？

GET my_index/_search/
{
    "query": {
        "match": {
           "_all": {
             "query": "Good and Bad",
             "analyzer": [
                 "stop_analyzer",
                 "synonym"
             ]
           }
        }
    }
}

上面的查询抛出一个错误:

{
   "error": {
      "root_cause": [
         {
            "type": "parsing_exception",
            "reason": "[match] unknown token [START_ARRAY] after [analyzer]",
            "line": 6,
            "col": 26
         }
      ],
      "type": "parsing_exception",
      "reason": "[match] unknown token [START_ARRAY] after [analyzer]",
      "line": 6,
      "col": 26
   },
   "status": 400
}

我想我不能在那里使用数组或对象，因为当我使用像"analyzer": "stop_analyzer"或"analyzer": "synonym"这样的单个分析器时，它的效果很好。所以我的问题是如何同时使用两者？

Answer 1

您可以定义一个custom analyzer，它可以将这两个简单的分析器组合为一个复杂的组件。

定义自定义分析器

假设您通过以下方式创建了索引:

PUT my_index
{  
  "settings": {
    "index": {
      "analysis": {
        "analyzer": {
          "stopwordsSynonym": {
            "filter": [
              "lowercase",
              "my_synonym",
              "english_stop"
            ],
            "tokenizer": "standard"
          }
        },
        "filter": {
          "english_stop": {
            "type": "stop",
            "stopwords": "_english_"
          },
          "my_synonym": {
            "type": "synonym",
            "synonyms": [
              "nice => good",
              "poor => bad"  
            ]
          }
        }
      }
    }
  },
  "mappings": {
    "my_type": {
        "properties": {
            "my_text": {
                "type": "text",
                "analyzer": "stopwordsSynonym"
            }
        }
    }
  }
}

并添加一条记录:

POST my_index/my_type
{
    "my_text": "People aren’t born good or bad. Maybe they’re born with tendencies either way, but it’s the way you live your life that matters."
}

现在默认情况下，对my_text的搜索将使用stopwordsSynonym分析器。该查询将与文档匹配，因为nice是good的同义词:

GET my_index/_search
{
    "query": {
        "match": {
            "my_text": "nice and ugly"
        }
    }
}

测试自定义分析仪

您也可以像这样测试分析仪:

GET my_index/_analyze 
{
  "analyzer": "stopwordsSynonym", 
  "text":     "nice or ugly"
}

{
   "tokens": [
      {
         "token": "good",
         "start_offset": 0,
         "end_offset": 4,
         "type": "SYNONYM",
         "position": 0
      },
      {
         "token": "ugly",
         "start_offset": 8,
         "end_offset": 12,
         "type": "<ALPHANUM>",
         "position": 2
      }
   ]
}

将此与standard分析器输出进行比较:

GET my_index/_analyze 
{
  "analyzer": "standard", 
  "text":     "nice or ugly"
}

{
   "tokens": [
      {
         "token": "nice",
         "start_offset": 0,
         "end_offset": 4,
         "type": "<ALPHANUM>",
         "position": 0
      },
      {
         "token": "or",
         "start_offset": 5,
         "end_offset": 7,
         "type": "<ALPHANUM>",
         "position": 1
      },
      {
         "token": "ugly",
         "start_offset": 8,
         "end_offset": 12,
         "type": "<ALPHANUM>",
         "position": 2
      }
   ]
}

实际上，stopwordsSynonym用nice替换了good token (其type是SYNONYM)，并从 token 列表中删除了or，因为它是常见的英语停用词。

定义分析器查询

为了对给定查询使用不同的分析器，可以使用 query_string 查询:

GET /_search
{
    "query": {
        "query_string": {
            "query": "my_text:nice and poor",
            "analyzer": "stopwordsSynonym"
        }
    }
}

或 match_phrase 查询:

GET my_index/_search
{
    "query": {
        "match_phrase" : {
            "my_standard_text" : {
                "query" : "nice and poor",
                "analyzer": "stopwordsSynonym"
            }
        }
    }
}

无论如何，应在创建时将analyzer添加到索引的设置中(请参阅答案的开头)。

还要看看search analyzer，它允许使用不同的分析器进行搜索。