ElasticSearch中Wildcard查询的Boost参数不生效问题求助

调整查询中wildcard部分的boost值时，结果得分完全没有变化。我们的搜索需求是：

1. 优先匹配title中的搜索短语
2. 其次匹配title和body中的内容
3. 若无结果则尝试模糊匹配
4. 同时通过时间衰减函数优先展示近期内容，30天内的信息权重远高于9个月前的

附上当前查询代码及_explain输出，求解决思路：

{
  "from": 0,
  "size": 10,
  "query": {
    "function_score": {
      "query": {
        "bool": {
          "should": [
            {
              "wildcard": {
                "title": {
                  "value": "*blackwell*",
                  "boost": 1000
                }
              }
            },
            {
              "multi_match": {
                "fields": [
                  "title",
                  "body"
                ],
                "query": "blackwell",
                "type": "phrase",
                "boost": 100
              }
            },
            {
              "multi_match": {
                "fields": [
                  "title",
                  "body"
                ],
                "query": "blackwell",
                "operator": "or",
                "boost": 10
              }
            },
            {
              "multi_match": {
                "fields": [
                  "title",
                  "body"
                ],
                "query": "blackwell",
                "fuzziness": "AUTO",
                "operator": "and",
                "boost": 1
              }
            }
          ],
          "minimum_should_match": 1
        }
      },
      "functions": [
        {
          "exp": {
            "modifiedAt": {
              "offset": "30d",
              "scale": "360d",
              "decay": 0.01
            }
          }
        }
      ],
      "boost_mode": "multiply"
    }
  }
}

_explain输出 #

{
  "_index": "534e9cac-96c9-47af-9321-67860f3b33ba_record_chunks",
  "_id": "836bba2c-8ba6-491e-a1e2-da7e91de78f8_0011",
  "matched": true,
  "explanation": {
    "value": 1254.3658,
    "description": "function score, product of:",
    "details": [
      {
        "value": 1254.3658,
        "description": "sum of:",
        "details": [
          {
            "value": 1254.3658,
            "description": "weight(body:blackwel in 257) [PerFieldSimilarity], result of:",
            "details": [
              {
                "value": 1254.3658,
                "description": "score(freq=2.0), computed as boost * idf * tf from:",
                "details": [
                  {
                    "value": 244.20001,
                    "description": "boost",
                    "details": []
                  },
                  {
                    "value": 7.447975,
                    "description": "idf, computed as log(1 + (N - n + 0.5) / (n + 0.5)) from:",
                    "details": [
                      {
                        "value": 544,
                        "description": "n, number of documents containing term",
                        "details": []
                      },
                      {
                        "value": 934570,
                        "description": "N, total number of documents with field",
                        "details": []
                      }
                    ]
                  },
                  {
                    "value": 0.68966836,
                    "description": "tf, computed as freq / (freq + k1 * (1 - b + b * dl / avgdl)) from:",
                    "details": [
                      {
                        "value": 2,
                        "description": "freq, occurrences of term within document",
                        "details": []
                      },
                      {
                        "value": 1.2,
                        "description": "k1, term saturation parameter",
                        "details": []
                      },
                      {
                        "value": 0.75,
                        "description": "b, length normalization parameter",
                        "details": []
                      },
                      {
                        "value": 56,
                        "description": "dl, length of field (approximate)",
                        "details": []
                      },
                      {
                        "value": 84.00777,
                        "description": "avgdl, average length of field",
                        "details": []
                      }
                    ]
                  }
                ]
              }
            ]
          }
        ]
      },
      {
        "value": 1,
        "description": "min of:",
        "details": [
          {
            "value": 1,
            "description": "Function for field modifiedAt:",
            "details": [
              {
                "value": 1,
                "description": "exp(- MIN[Math.max(Math.abs(1.731852097E12(=doc value) - 1.732904021506E12(=origin))) - 2.592E9(=offset), 0)] * 1.4805716904539902E-10)",
                "details": []
              }
            ]
          },
          {
            "value": 3.4028235e38,
            "description": "maxBoost",
            "details": []
          }
        ]
      }
    ]
  }
}

• 先确认wildcard是否命中目标文档：从_explain输出能看到，当前文档只匹配了body字段的multi_match查询，wildcard查询根本没触发。调整boost值自然不会影响得分——先找一个title包含blackwell的文档，用_explain验证wildcard是否命中，再测试boost是否生效。
• 替换wildcard为更高效的精确匹配：*blackwell*这种前后通配符的查询无法利用倒排索引，是全字段扫描，而且默认是固定常数得分（仅乘以boost）。如果要优先匹配title中的短语，直接用match_phrase针对title字段，boost设为1000，比wildcard更高效，也更贴合“优先匹配title短语”的需求。
• 用dis_max替代bool should实现优先级：当前bool should是叠加所有匹配查询的得分，可能导致低权重查询的得分叠加后覆盖高权重查询的优先级。改用dis_max查询，会取匹配查询中的最高分，再加上其他查询的tie_breaker分数，能确保高boost的查询（比如title短语）成为得分主导。示例结构：

  "dis_max": {
    "queries": [
      {
        "match_phrase": {
          "title": {
            "query": "blackwell",
            "boost": 1000
          }
        }
      },
      {
        "multi_match": {
          "fields": ["title", "body"],
          "query": "blackwell",
          "type": "phrase",
          "boost": 100
        }
      },
      // 其余查询...
    ],
    "tie_breaker": 0.1
  }
  

• 优化时间衰减函数参数：当前文档的时间衰减值为1，说明它在30天的offset范围内，所以时间函数没起作用。如果要让30天内的文档权重显著高于旧文档，可以把scale改为30d，decay设为0.5，这样超过30天后得分会快速衰减，更符合需求。
• 排查wildcard boost不生效的其他情况：如果确实有文档命中wildcard但boost无效，检查字段的index_options是否正确，或者是否有自定义计分器覆盖了boost。另外，wildcard的基础得分是固定值（默认1），boost是乘在这个值上，如果multi_match的tf/idf得分本身就比wildcard的boost后得分高，叠加后可能不明显，这时候用dis_max就能解决问题。

内容的提问来源于stack exchange，提问作者JackBurton