arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2611.06662cs.AI
收藏

搜索智能体为何出错:基于图的步骤级验证

Where Search Agents Go Off the Rails: Graph-Grounded Step-Level Verification

Seoul National University(首尔大学);Korea University(高丽大学)

Hyeseo Jeon, Seonho Yoo, Jay-Yoon Lee

AI总结:

针对搜索增强推理系统中间步骤出错的问题,提出GraphRail方法,通过结构化主张比较实现步骤级验证,在多跳QA基准和推理框架上提升错误召回率,还可指导在线推理修正中间错误。

摘要:

搜索增强推理系统能得出正确答案,但中间步骤可能出错。可靠的审计需检查证据D是否覆盖问题Q、推理T是否针对Q、T是否基于D。我们提出GraphRail,它将需求、证据和推理表示为结构化主张,无需生成式判断或标准答案即可比较。在三个多跳QA基准和三个推理框架上,GraphRail的错误召回率优于生成式验证器,且在随机提取下保持稳定;其诊断还可指导在线推理,提供可追溯的中间错误检测与修正方式。

英文摘要:

Search-augmented reasoning systems can reach correct answers while making errors in intermediate steps. Reliable auditing must check whether evidence $D$ covers the question $Q$, reasoning $T$ addresses $Q$, and $T$ is grounded in $D$. We introduce GraphRail, which represents requirements, evidence, and reasoning as structured claims and compares them without a generative judge or gold answer. Across three multi-hop QA benchmarks and three reasoning frameworks, GraphRail improves error recall over generative verifiers while remaining stable under stochastic extraction. Its diagnoses can also guide online reasoning, providing a traceable way to detect and correct intermediate errors.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑