The requested URL was not found on this server. 不管你信不信,反正我是没找到

分词匹对,分值问题,如何提高精确度

Elasticsearch | 作者 zp9755 | 发布于2021年12月14日 | 阅读数:928

使用ik_smart,用matchQuery查询盛和小区2-201,得到结果如下
{
"score": 12.894995,
"createTime": "2021-12-13T08:11:01.625Z",
"xqmc": "寺巷镇集镇2-201号",
"pageSize": 10,
"sqbm": "321203006003",
"pageNum": 1,
"addressId": "257546"
},
{
"score": 5.7983465,
"createTime": "2021-12-13T08:10:22.658Z",
"xqmc": "凤凰东路99号盛和花园33幢",
"pageSize": 10,
"sqbm": "321203004012",
"pageNum": 1,
"addressId": "144808"
}
如何做能将第二个结果放在前面?
已邀请:

hapjin

赞同来自:

ES 作为召回,ES的打分作为粗排分数。将召回结果返回给 rank 服务,在 rank 里面实现各种排序干预逻辑。:)

xiaowuge - 90后

赞同来自:

索引创建方法(一个分片)
PUT /test_score
{
  "settings" : {
        "number_of_shards" : 1
    },
  "mappings": {
    "_doc": {
      "properties": {
        "xqmc":{
          "type":"text","analyzer":"ik_smart"
        }
      }
    }
  }
}
 
用ik_smart解析词条:
GET /test_score/_analyze
{
  "analyzer": "ik_smart"
  , "text": ["寺巷镇集镇2-201"]
}
 
{
  "tokens": [
    {
      "token": "寺",
      "start_offset": 0,
      "end_offset": 1,
      "type": "CN_CHAR",
      "position": 0
    },
    {
      "token": "巷",
      "start_offset": 1,
      "end_offset": 2,
      "type": "CN_CHAR",
      "position": 1
    },
    {
      "token": "镇",
      "start_offset": 2,
      "end_offset": 3,
      "type": "CN_CHAR",
      "position": 2
    },
    {
      "token": "集镇",
      "start_offset": 3,
      "end_offset": 5,
      "type": "CN_WORD",
      "position": 3
    },
    {
      "token": "2-201",
      "start_offset": 5,
      "end_offset": 10,
      "type": "LETTER",
      "position": 4
    }
  ]
}
 
用ik_smart解析词条二:
GET /test_score/_analyze
{
  "analyzer": "ik_smart"
  , "text": ["凤凰东路99号盛和花园33幢"]
}
 
 
{
  "tokens": [
    {
      "token": "凤凰",
      "start_offset": 0,
      "end_offset": 2,
      "type": "CN_WORD",
      "position": 0
    },
    {
      "token": "东路",
      "start_offset": 2,
      "end_offset": 4,
      "type": "CN_WORD",
      "position": 1
    },
    {
      "token": "99号",
      "start_offset": 4,
      "end_offset": 7,
      "type": "TYPE_CQUAN",
      "position": 2
    },
    {
      "token": "盛",
      "start_offset": 7,
      "end_offset": 8,
      "type": "CN_CHAR",
      "position": 3
    },
    {
      "token": "和",
      "start_offset": 8,
      "end_offset": 9,
      "type": "CN_CHAR",
      "position": 4
    },
    {
      "token": "花园",
      "start_offset": 9,
      "end_offset": 11,
      "type": "CN_WORD",
      "position": 5
    },
    {
      "token": "33幢",
      "start_offset": 11,
      "end_offset": 14,
      "type": "TYPE_CQUAN",
      "position": 6
    }
  ]
}
 
用ik_smart解析查询词条三:
GET /test_score/_analyze
{
  "analyzer": "ik_smart"
  , "text": ["盛和小区2-201"]
}
 
{
  "tokens": [
    {
      "token": "盛",
      "start_offset": 0,
      "end_offset": 1,
      "type": "CN_CHAR",
      "position": 0
    },
    {
      "token": "和",
      "start_offset": 1,
      "end_offset": 2,
      "type": "CN_CHAR",
      "position": 1
    },
    {
      "token": "小区",
      "start_offset": 2,
      "end_offset": 4,
      "type": "CN_WORD",
      "position": 2
    },
    {
      "token": "2-201",
      "start_offset": 4,
      "end_offset": 9,
      "type": "LETTER",
      "position": 3
    }
  ]
}
 
从解析结果中可以看出,二和三会有两个词条匹配上:"token": "盛"    "token": "和",而一和三只有"token": "2-201" 匹配上;为此二的分数肯定高!
 
最后直接查询:
GET /test_score/_search
{
  "query": {
    "match": {
      "xqmc": "盛和小区2-201"
    }
  }
}
 
结果为:
{
  "took": 3,
  "timed_out": false,
  "_shards": {
    "total": 1,
    "successful": 1,
    "skipped": 0,
    "failed": 0
  },
  "hits": {
    "total": 2,
    "max_score": 1.3440006,
    "hits": [
      {
        "_index": "test_score",
        "_type": "_doc",
        "_id": "iXzdyn0BRxwSc_uunfyp",
        "_score": 1.3440006,
        "_source": {
          "xqmc": "凤凰东路99号盛和花园33幢"
        }
      },
      {
        "_index": "test_score",
        "_type": "_doc",
        "_id": "iHzdyn0BRxwSc_uukvym",
        "_score": 0.7156682,
        "_source": {
          "xqmc": "寺巷镇集镇2-201号"
        }
      }
    ]
  }
}
 
所以,初步猜测你可能是分片原因导致排序问题,你可以在查询语句后面加上 "explain": true,分析一下打分,类似这样:
GET /test_score/_search
{
  "query": {
    "match": {
      "xqmc": "盛和小区2-201"
    }
  }
  , "explain": true
}
 
 

要回复问题请先登录注册