Elasticsearch数组字段高频兴趣统计：聚合查询失效问题求助

索引数据 #

{"index":{"_index":"companydatabase"}}  
 {"FirstName":"ELVA","LastName":"RECHKEMMER","Designation":"CEO","Salary":"154000","DateOfJoining":"1993-01-11","Address":"8417 Blue Spring St. Port Orange, FL 32127","Gender":"Female","Age":62,"MaritalStatus":"Unmarried","Interests":["Body Building","Illusion","Protesting","Taxidermy","TV watching","Cartooning","Skateboarding"]}
{"index":{"_index":"companydatabase"}}  
 {"FirstName":"JENNEFER","LastName":"WENIG","Designation":"President","Salary":"110000","DateOfJoining":"2013-02-07","Address":"16 Manor Station Court Huntsville, AL 35803","Gender":"Female","Age":45,"MaritalStatus":"Unmarried","Interests":["String Figures","Working on cars","Button Collecting","Surf Fishing"]}
{"index":{"_index":"companydatabase"}}

问题原因 #

你的聚合请求失效的核心问题是terms聚合的排序字段错误：Elasticsearch的terms聚合默认统计的计数字段是_count，而非你写的count，这会导致排序逻辑无法识别，进而无法正确返回兴趣的统计结果。

修正后的请求体与代码 #

修正后的请求体 #

request_body = {
  "size": 0,
  "aggs": {
    "interests": {
      "terms": {
        "field": "Interests.keyword",
        "size": 10,
        "order": {
          "_count": "desc"  # 修正为正确的计数字段名
        }
      }
    }
  }
}

正确调用方式 #

无需额外包裹JSON()，直接调用es.search即可获取聚合结果：

response = es.search(index="companydatabase", body=request_body)
# 打印聚合结果
print(response['aggregations']['interests']['buckets'])

结果说明 #

修正后，请求会返回按兴趣出现次数降序排列的桶（bucket），每个桶包含兴趣名称（key）和对应的出现次数（doc_count），你可以从中直接获取最常见的兴趣。

内容的提问来源于stack exchange，提问作者David Vazquez