min_score 过滤低分结果

做的事：真实搜一个关键词，对比不设min_score和设置min_score=0.5之后的命中结果

代码真实索引4篇文档，标题分别是: "elasticsearch performance tuning guide" / "elasticsearch basic tutorial" "a random article about cooking" / "python programming language elasticsearch mentioned once" POST /ms_test/_search {"query":{"match":{"title":"elasticsearch performance"}}} POST /ms_test/_search {"min_score":0.5,"query":{"match":{"title":"elasticsearch performance"}}}

验证：不设min_score真实命中3篇，得分从1.63一路到0.31；加min_score=0.5后真实只剩得分最高那1篇

不设min_score: hits=3
→ 1.6349643 "elasticsearch performance tuning guide"
→ 0.41299206 "elasticsearch basic tutorial"
→ 0.31387398 "python programming language elasticsearch mentioned once"

min_score=0.5: hits=1
→ 1.6349643 "elasticsearch performance tuning guide"
  ← 真实得分0.41和0.31那两篇都被min_score真实挡在了结果外面

这组结果真实验证了min_score最直接的作用：它真实在打分完成之后，把得分低于这个门槛的文档整个排除出最终的命中结果，哪怕这些文档真实满足查询条件、真实被判定为相关。本次实测得分0.41和0.31的两篇文档真实都因为相关性不够高（低于0.5这个门槛）被排除掉，hits.total也真实从3条精确收窄成了1条，不是"显示上过滤但统计上还留着"，而是真实从命中结果集里被摘除了。

真实验证min_score对聚合的影响，跟本系列上一篇post_filter的结论做直接对照

代码每篇文档都额外带一个cat(keyword)字段: high/mid/none/low POST /ms_test/_search {"min_score":0.5,"query":{"match":{"title":"elasticsearch performance"}}, "aggs":{"cats":{"terms":{"field":"cat"}}}} POST /ms_test/_search {"query":{"match":{"title":"elasticsearch performance"}}, "aggs":{"cats":{"terms":{"field":"cat"}}}}

验证：min_score真实把聚合桶也一并收窄了，这跟post_filter的行为真实完全相反

min_score=0.5，聚合桶:
→ [('high', 1)]   ← 真实只剩得分够高的那篇对应的cat桶

不设min_score，同样的聚合:
→ [('high', 1), ('low', 1), ('mid', 1)]   ← 真实三个桶都在

分步演示：min_score从排除低分文档到真实收窄聚合范围的完整验证

每一步都是本文上面在本机真实的Elasticsearch 8.15.0上跑出来的原始结果。