使用模糊滤波器进行聚合

Question

Elastisearch中是否有可能有一个包含模糊的过滤器/查询的聚合？

ATM我有包含嵌套对象[]的文档。我想要实现的目标：

- select from each document 0..n nested objects which match a filter
- from this array of nested objects take the distinct one
- sort them by _score
- take the top 5 or X
- use the terms for an autocomplete/suggestions (should work more as a "like" and not autocomplete)

到目前为止，我尝试了不同类型的聚合，例如：significant_terms，top_hits但没有很好的组合，所以我没有得到理想的结果。 问题：

1. significant_terms不会返回一个值，直到他找出一个术语很重要的时候（也许我没有使用好的分析器）
2. top-hits从所选文档返回任何嵌套的obj，并且还包含重复项

以下是我的查询示例

  GET customerinsights/_search

    {
  "query": {
    "bool": {
      "must": [
        {
          "nested": {
            "path": "CustomerInsightTargets",
            "query": {
              "bool": {
                "must": [
                  {
                    "match": {
                      "CustomerInsightTargets.CustomerInsightValue": {
                        "query": "2017",
                        "operator": "AND",
                        "fuzziness": 2
                      }
                    }
                  }
                ]
              }
            }

          }
        }
      ]
    }
  } ,
  "aggs": {
    "root": {
      "nested": {
        "path": "CustomerInsightTargets"
      },
      "aggs": {
        "top_tags": {
          "terms": {
            "field": "CustomerInsightTargets.CustomerInsightSource.keyword"
          },
          "aggs": {
            "top_tag_hits": {
              "top_hits": {
                "sort": [
                  {
                    "_score": {
                      "order": "desc"
                    }
                  }
                ],
                "size": 5,
                "_source": "CustomerInsightTargets"
              }
            }
          }
        }
      }
    }
  },
  "size": 0,
  "_source": "CustomerInsightTargets"
}