Elasticsearch 查询以查找丢失的记录

Question

我一直在尝试使用 DSL 寻找一种方法来查找一组文档中缺少的文档。在我的数据集中，我有：

Unique ID | Information
abc       | Some data
abc       | Special Information
abc       | Some data
def       | Some data
def       | Special Information
def       | Some data
ghi       | Some data
ghi       | Some data

我想设计一个查询，为我提供没有特殊信息

的文档集的 UniqueID

例如，对于上述数据集，结果将是 ghi

谢谢

Answer 1

有多个没有特殊信息的唯一 ID。从这里开始并根据需要进行调整：

设置

PUT special_info
{
  "mappings": {
    "properties": {
      "unique_id": {
        "type": "keyword"
      },
      "information": {
        "type": "text",
        "fields": {
          "keyword": {
            "type": "keyword"
          }
        }
      }
    }
  }
}

同步

POST _bulk
{"index":{"_index":"special_info","_type":"_doc"}}
{"unique_id":"abc","information":"Some data"}
{"index":{"_index":"special_info","_type":"_doc"}}
{"unique_id":"abc","information":"Special Information"}
{"index":{"_index":"special_info","_type":"_doc"}}
{"unique_id":"abc","information":"Some data"}
{"index":{"_index":"special_info","_type":"_doc"}}
{"unique_id":"def","information":"Some data"}
{"index":{"_index":"special_info","_type":"_doc"}}
{"unique_id":"def","information":"Special Information"}
{"index":{"_index":"special_info","_type":"_doc"}}
{"unique_id":"def","information":"Some data"}
{"index":{"_index":"special_info","_type":"_doc"}}
{"unique_id":"ghi","information":"Some data"}
{"index":{"_index":"special_info","_type":"_doc"}}
{"unique_id":"ghi","information":"Some data"}

查询

GET special_info/_search
{
  "query": {
    "bool": {
      "must_not": [
        {
          "term": {
            "information.keyword": {
              "value": "Special Information"
            }
          }
        }
      ]
    }
  },
  "_source": "unique_id",
  "aggs": {
    "by_unique_ids": {
      "terms": {
        "field": "unique_id"
      }
    }
  }
}

屈服

...
"aggregations" : {
    "by_unique_ids" : {
      "doc_count_error_upper_bound" : 0,
      "sum_other_doc_count" : 0,
      "buckets" : [
        {
          "key" : "abc",
          "doc_count" : 2
        },
        {
          "key" : "def",
          "doc_count" : 2
        },
        {
          "key" : "ghi",
          "doc_count" : 2
        }
      ]
    }
  }

Answer 2

上面我是用聚合来解决的

我用过terms aggregation, filter aggregation and bucket selector aggregation

使用术语聚合创建 unique_id 的存储桶。获取具有特殊信息的术语下的文档数。如果 count==0 那么 return bucket.

查询：

{
  "size": 0,
  "aggs": {
    "unique_id": {
      "terms": {
        "field": "unique_id",
        "size": 10
      },
      "aggs": {
        "filter_special_infor": {
          "filter": {
            "term": {
              "information.keyword": "Special Information"
            }
          },
          "aggs": {
            "filtered_count": {
              "value_count": {
                "field": "unique_id"
              }
            }
          }
        },
        "doc_with_no_special_infor": {
          "bucket_selector": {
            "buckets_path": {
              "filteredCount": "filter_special_infor>filtered_count"
            },
            "script": "if(params.filteredCount==0){return true;}else{return false;}"
          }
        }
      }
    }
  }
}

结果：

"unique_id" : {
      "doc_count_error_upper_bound" : 0,
      "sum_other_doc_count" : 0,
      "buckets" : [
        {
          "key" : "ghi",
          "doc_count" : 2,
          "filter_special_infor" : {
            "doc_count" : 0,
            "filtered_count" : {
              "value" : 0
            }
          }
        }
      ]
    }

Elasticsearch 查询以查找丢失的记录

Elasticsearch query to find missing records

elasticsearch

dsl

elasticsearch-aggregation

elasticsearch-query