Skip to content

fix/短查询命中验证修复#22

Merged
Snailclimb merged 1 commit intoSnailclimb:masterfrom
YNAlone:fix/短查询命中验证修复
Apr 7, 2026

Hidden character warning

The head ref may contain hidden characters: "fix/\u77ed\u67e5\u8be2\u547d\u4e2d\u9a8c\u8bc1\u4fee\u590d"
Merged

fix/短查询命中验证修复#22
Snailclimb merged 1 commit intoSnailclimb:masterfrom
YNAlone:fix/短查询命中验证修复

Conversation

@YNAlone
Copy link
Copy Markdown
Contributor

@YNAlone YNAlone commented Apr 4, 2026

问题

用户提问“什么是进程”等中文语句,向量检索显示成功检索到12条相关文档,但是RAG仍然返回“未检索到相关信息”

原因

hasEffectiveHit 的短 token 命中确认逻辑存在两处缺陷:

  1. SHORT_TOKEN_PATTERN^[\p{L}\p{N}_-]{2,20}$)中 \p{L} 包含中文字符,
    导致 Query 改写后的长短语(如"操作系统中进程的定义与基本概念",15字无空格)也被判定为短 token,触发字面匹配并失败。
  2. 对"什么是进程"做字面匹配时,搜索的是完整问句,而知识库文档只包含"进程",导致匹配失败。

修复

  • isShortTokenQuery:含 CJK 字符时长度限制收紧为 ≤6
  • hasEffectiveHit:新增 extractCoreTerm 提取中文问句核心词再匹配

额外修复

Redisson 4.0.0 的 readGroup 在无消息时返回 EmptyList 而非空 Map,
内部强转失败抛出 ClassCastException,导致所有 Stream 消费者持续刷错误日志。
RedisService.streamConsumeMessages 捕获该异常并静默处理。

@Snailclimb Snailclimb merged commit 765a638 into Snailclimb:master Apr 7, 2026
@Snailclimb Snailclimb added the enhancement New feature or request label Apr 7, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

enhancement New feature or request

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants