弹性搜索 N-gram 未返回预期结果

Question

试图计算出这个微不足道的例子的得分。我希望得到 brenda eaton 的文档，但我得到 brenda fassie 作为最好的结果。

PUT ngram
{
  "settings": {
    "analysis": {
      "analyzer": {
        "my_analyzer": {
          "tokenizer": "my_tokenizer"
        }
      },
      "tokenizer": {
        "my_tokenizer": {
          "type": "ngram",
          "min_gram": 3,
          "max_gram": 3,
          "token_chars": [
            "letter",
            "digit"
          ]
        }
      }
    }
  },
   "mappings": {
    "tweet" : {
      "properties" : {
        "text" : {
          "type" :    "text",
          "analyzer": "my_analyzer"
        }
      }
    }
  }
}

PUT ngram/tweet/1
{
 "text":"searched the blue sky during the summer"
}
PUT ngram/tweet/2
{
 "text":"sdssded the trans hex during the sssss"
}

PUT ngram/tweet/3
{
 "text":"searched the brenda eaton during the summer"
}
PUT ngram/tweet/4
{
 "text":"sdssded the brenda fassie during the sssss"
}


GET ngram/_search
{
    "query": {
        "match" : {
            "text" : {
                "query" : "brenda eaton",
                "max_expansions" : 10
            }
        }
    }
}

Answer 1

在填充索引的初始阶段，文档的相关性可能在很大程度上取决于它们在分片中的分布。尝试用一个主分片和一个副本分片创建索引，你会得到想要的结果。

您可以在 Elasticsearch 指南的以下文章中找到对这种现象的很好解释：Relevance is broken!

弹性搜索 N-gram 未返回预期结果

Elastic search N-gram not returning expected result

elasticsearch

kibana