完成建议者显示错误结果

Question

我正在使用Elasticsearch。我使用下面给出的映射创建了索引自动完成功能。我正在尝试多个基于上下文的自动完成。当我检查结果时，发现建议不正确。

我有一个层次自动完成的情况。 我是第一位成为独立自动完成领域的经销商。现场状态将经销商作为自动完成建议的上下文。现场县有经销商和州作为自动完成领域的上下文。

mapping ={
        "mappings": {
            "panels": {
                "properties": {
                    "dealer": {
                        "type": "completion"
                    },
                    "county": {
                        "type": "completion",
                        "contexts": [{
                                "name": "dealername",
                                "type": "category",
                                "path": "dealername"
                            },
                            {
                                "name": "statename",
                                "type": "category",
                                "path": "statename"
                            }
                        ]
                    },
                    "state": {
                        "type": "completion",
                        "contexts": [{
                            "name": "dealername",
                            "type": "category",
                            "path": "dealername"
                        }]
                    },
                    "dealername":{
                        "type": "keyword"
                    },
                    "statename":{
                        "type": "keyword"
                    },
                    "countyname":{
                        "type": "keyword"
                    },
                    "meid": {
                        "type": "text"
                    }
                }
            }
        }
    }
    res = es.indices.create(index="autocomplete", ignore=[400], body=mapping)
    print(res)

    def insertstatewise():
        rows = session.execute("SELECT dealer_name, statename, county_name, meid from analytics.nearest_reco_output_1")
        i = 0
        for row in rows:
            i += 1
            yield {
                "_id": row["meid"],
                "_index": "autocomplete",
                "_type": 'panels',
                "_source":
                    {
                    "dealer": {
                    "input": [row["dealer_name"]]
                    },
                    "state": {
                    "input": [row["statename"]],
                    "contexts": {
                        "dealername": [row["dealer_name"]]
                        }
                    },
                    "county": {
                    "input": [row["county_name"]],
                    "contexts": {
                        "dealername": [row["dealer_name"]],
                        "statename": [row["statename"]]
                        }
                    },
                    "dealername": row["dealer_name"],
                    "statename": row["statename"],
                    "countyname": row["county_name"],
                    "meid": row["meid"]
                    },
            }
    xp = bulk(es, insertstatewise())
    print(xp)

当我使用curl请求查询索引时

curl -X POST "########:9200/autocomplete/_search?pretty" -H 'Content-Type: application/json' -d'
    {
        "suggest": {
            "state_suggestion_for_dealer" : {
                "prefix" : "dealer2",
                "completion" : {
                    "field" : "dealer",
                    "size": 10
                }
            }
        }
    }

预期结果应为delaer2

对于完成样本，我得到了错误的结果

{
    "text" : "dealer3",
    "_index" : "autocomplete",
    "_type" : "panels",
    "_id" : "#############",
    "_score" : 1.0,
    "_source" : {
      "dealer" : {
        "input" : [
          "dealer3"
        ]
      },
      "state" : {
        "input" : [
          "Virginia"
        ],
        "contexts" : {
          "dealername" : [
            "dealer3"
          ]
        }
      },
      "county" : {
        "input" : [
          "Park"
        ],
        "contexts" : {
          "dealername" : [
            "dealer3"
          ],
          "statename" : [
            "Virginia"
          ]
        }
      },
      "dealername" : "dealer3",
      "statename" : "Virginia",
      "countyname" : "Park",
      "meid" : "#########"
    }
   }

正如我查询的前缀为Dealer2的完成那样，那么为什么完成建议中包含有Dealer3文档。