elasticsearch completion suggest on multifield

Question

我正在尝试从多领域获得建议。我找不到这样的示例，所以这可能不是最好的主意，但我对您的意见很感兴趣。

映射:

POST /authors
    {
       "mappings": {
          "author": {
             "properties": {
                "name": {
                   "type": "multi_field",
                   "fields": {
                      "name": {
                         "type": "string",
                         "index": "analyzed"
                      },
                      "ac": {
                         "type": "completion",
                         "index_analyzer": "simple",
                         "search_analyzer": "simple",
                         "payloads": true
                      }
                   }
                }
             }
          }
       }
    }

数据:

POST /authors/author/1
    {
       "name": "Fyodor Dostoevsky"
    }

查询:

POST /authors/_suggest

    {
       "authorsAutocomplete": {
          "text": "fyodor",
          "completion": {
             "field": "name.ac"
          }
       }
    }

要求是:

• get 查询适用于文本“fyodor”，也适用于“dostoevsky”，此示例仅适用于“fyodor”
• 能够过滤建议

有什么想法可以实现这些吗？

Answer 1

首先，suggesters 在多领域效果不佳，所以你可能想把它放在外面。 其次，要让您同时使用名字和名字进行查询，您必须在索引数据时选择输出/输入。

SENSE 的工作代码示例:

POST authors

PUT authors/_mapping/author
{
    "properties" : {
        "name" : { "type" : "string" },
        "suggest" : { "type" : "completion"}
    }
}

POST authors/author/1
{
    "name": "Fyodor Dostoevsky",
    "suggest": {
        "input": ["Dostoevsky", "Fyodor"],
        "output": "Fyodor Dostoevsky"
    }
}

POST authors/_suggest
{
    "authorsAutocomplete": {
        "text": "d",
        "completion": {
            "field": "suggest"
        }
    }
}

DELETE authors

结果:

{
    "_shards": {
        "total": 5,
        "successful": 5,
        "failed": 0
    },
    "authorsAutocomplete": [
        {
            "text": "d",
            "offset": 0,
            "length": 1,
            "options": [
                {
                    "text": "Fyodor Dostoevsky",
                    "score": 1
                }
            ]
        }
    ]
}

过滤器不适用于建议。要实现某种过滤，你可以看看这个 blog post关于在建议中使用上下文。